Download Candidate gene scan for Single Nucleotide Polymorphisms involved

bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1
2
3
4
Title: Candidate gene scan for Single Nucleotide
Polymorphisms involved in the determination of
normal
variability
in
human
craniofacial
morphology
5
6
7
8
Authors: Mark Barash1,2*, Philipp E. Bayer3,4, Angela van Daal2
9
10
11
12
13
14
15
16
Affiliations:
1
School of Mathematical and Physical Sciences, Centre for Forensic Sciences, Faculty of
Science, University of Technology Sydney, Sydney, NSW, Australia
2
Faculty of Health Sciences and Medicine, Bond University, Gold Coast, QLD, Australia
3
School of Agriculture and Food Sciences, The University of Queensland, Brisbane, QLD,
Australia
4
Australian Centre for Plant Functional Genomics, The University of Queensland, Brisbane,
17
QLD, Australia
18
Emails: Mark Barash [email protected]; Philipp E. Bayer
19
[email protected]; Angela van Daal [email protected]
20
∗ Corresponding author email: [email protected]
21
∗ Corresponding author postal address: Centre for Forensic Sciences, School of
22
Mathematical and Physical Sciences, Faculty of Science University of Technology
23
Sydney, Building 4, Thomas St, Broadway NSW 2007, Australia.
24
25
26
Page 1
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
27
Abstract
28
Despite intensive research on genetics of the craniofacial morphology using animal models
29
and human craniofacial syndromes, the genetic variation that underpins normal human facial
30
appearance is still largely elusive. Recent development of novel digital methods for capturing
31
the complexity of craniofacial morphology in conjunction with high-throughput genotyping
32
methods, show great promise for unravelling the genetic basis of such a complex trait.
33
As a part of our efforts on detecting genomic variants affecting normal craniofacial
34
appearance, we have implemented a candidate gene approach by selecting 1,201 single
35
nucleotide polymorphisms (SNPs) and 4,732 tag SNPs in over 170 candidate genes and
36
intergenic regions. We used 3-dimentional (3D) facial scans and direct cranial measurements
37
of 587 volunteers to calculate 104 craniofacial phenotypes. Following genotyping by
38
massively parallel sequencing, genetic associations between 2,332 genetic markers and 104
39
craniofacial phenotypes were tested.
40
An application of a Bonferroni–corrected genome–wide significance threshold produced
41
significant associations between five craniofacial traits and six SNPs. Specifically,
42
associations of nasal width with rs8035124 (15q26.1), cephalic index with rs16830498
43
(2q23.3), nasal index with rs37369 (5q13.2), transverse nasal prominence angle with
44
rs59037879 (10p11.23) and rs10512572 (17q24.3), and principal component explaining
45
73.3% of all the craniofacial phenotypes, with rs37369 (5p13.2) and rs390345 (14q31.3) were
46
observed.
47
Due to over-conservative nature of the Bonferroni correction, we also report all the
48
associations that reached the traditional genome-wide p-value threshold (<5.00E-08) as
49
suggestive. Based on the genome-wide threshold, 8 craniofacial phenotypes demonstrated
50
significant associations with 34 intergenic and extragenic SNPs. The majority of associations
51
are novel, except PAX3 and COL11A1 genes, which were previously reported to affect
52
normal craniofacial variation.
53
This study identified the largest number of genetic variants associated with normal variation
54
of craniofacial morphology to date by using a candidate gene approach, including
55
confirmation of the two previously reported genes. These results enhance our understanding
56
of the genetics that determines normal variation in craniofacial morphology and will be of
57
particular value in medical and forensic fields.
Page 2
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
58
59
Keywords
60
SNPs, single nucleotide polymorphisms, craniofacial, facial appearance, embryogenetics
61
forensic DNA phenotyping, facial reconstruction.
62
63
Author Summary
64
There is a remarkable variety of human facial appearances, almost exclusively the result of
65
genetic differences, as exemplified by the striking resemblance of identical twins. However,
66
the genes and specific genetic variants that affect the size and shape of the cranium and the
67
soft facial tissue features are largely unknown. Numerous studies on animal models and
68
human craniofacial disorders have identified a large number of genes, which may regulate
69
normal craniofacial embryonic development.
70
In this study we implemented a targeted candidate gene approach to select more than 1,200
71
polymorphisms in over 170 genes that are likely to be involved in craniofacial development
72
and morphology. These markers were genotyped in 587 DNA samples using massively
73
parallel sequencing and analysed for association with 104 traits generated from 3-
74
dimensional facial images and direct craniofacial measurements. Genetic associations (p-
75
values<5.00E-08) were observed between 8 craniofacial traits and 34 single nucleotide
76
polymorphisms (SNPs), including two previously described genes and 26 novel candidate
77
genes and intergenic regions. This comprehensive candidate gene study has uncovered the
78
largest number of novel genetic variants affecting normal facial appearance to date. These
79
results will appreciably extend our understanding of the normal and abnormal embryonic
80
development and impact our ability to predict the appearance of an individual from a DNA
81
sample in forensic criminal investigations and missing person cases.
82
83
Introduction
84
The human face is probably the most commonly used descriptor of a person and has
85
an extraordinary role in human evolution, social interactions, clinical applications as well as
86
forensic investigations. The influence of genes on facial appearance can be seen in the
Page 3
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
87
striking resemblance of monozygotic twins as well as amongst first degree relatives,
88
indicating a high heritability [1, 2].
89
Uncovering the genetic background for regulation of craniofacial morphology is not a trivial
90
task. Human craniofacial development is a complex multistep process, involving numerous
91
signalling cascades of factors that control neural crest development, followed by a number of
92
epithelial-mesenchymal interactions that control outgrowth, patterning and skeletal
93
differentiation, as reviewed by Sperber et. al. [2]. The mechanisms involved in this process
94
include various gene expression and protein translation patterns, which regulate cell
95
migration, positioning and selective apoptosis, subsequently leading to development of
96
specific facial prominences. These events are precisely timed and are under hormonal and
97
metabolic control. Most facial features of the human embryo are recognizable from as early
98
as 6 weeks post conception, developing rapidly in utero and continuing to develop during
99
childhood and adolescence [3, 4]. Development of the face and brain are interconnected and
100
occur at the same time as limb formation. Facial malformations therefore, frequently occur
101
with brain and limb abnormalities and vice versa. Genetic regulation of craniofacial
102
development involves several key morphogenic factors such as HOX, WNT, BMP, FGF as
103
well as hundreds of other genes and intergenic regulatory regions, incorporating numerous
104
polymorphisms [2]. The SNPs involved in craniofacial diseases may in fact influence the
105
extraordinary variety of human facial appearances, in the same way that genes responsible for
106
albinism have been shown to be involved in normal pigmentation phenotypes [5].
107
Additionally, non-genetic components such as nutrition, climate and socio-economic
108
environment may also affect human facial morphology via epigenetic regulation of
109
transcription, translation and other cellular mechanics. To date, both the genetic and even
110
more so, the epigenetic regulation of craniofacial morphology shaping are poorly understood.
111
The genetic basis of craniofacial morphogenesis has been explored in numerous animal
112
models with multiple loci shown to be involved [2]. The majority of human studies in this
113
field have focused on the genetics of various craniofacial disorders such as craniosynostosis
114
and cleft lip/palate [6, 7], which may provide a link to regulation of normal variation of the
115
craniofacial phenotype, as for example observed between cleft-affected offspring and the
116
increase of facial width seen in non-affected parents [8]. These studies have identified several
117
genes with numerous genetic variants that may contribute to normal variation of different
118
facial features, such as cephalic index, bizygomatic distance and nasal area measurements [9-
119
11]. Studies of other congenital disorders involving manifestation of craniofacial
Page 4
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
120
abnormalities such as Alagille syndrome (JAG1 and NOTCH2 gene mutations), Down
121
syndrome (chromosome 21 trisomy - multiple genes), Floating-Harbor syndrome (SRCAP
122
gene mutations) and Noonan syndrome (mutations in various genes such as PTPN11 and
123
RAF1) provide additional information on the candidate genes potentially involved in normal
124
craniofacial development [12-17].
125
In recent years, new digital technologies such as 3-Dimentional laser imaging have been used
126
in numerous anthropometric studies. 3-D laser imaging allows accurate and rapid capture of
127
facial morphology, providing a better alternative to traditional manual measurements of
128
craniofacial distances [18-20]. The high-throughput genotyping technologies and digital
129
methods for capturing facial morphology have been used in a number of recent studies that
130
demonstrated a link between normal facial variation and specific genetic polymorphisms [21-
131
23]. Despite these promising results, our current knowledge of craniofacial genetics is sparse.
132
This study aims to further define the polymorphisms associated with normal facial variation
133
using a candidate gene approach. The advantage of a candidate gene approach over previous
134
genome wide association studies (GWAS) is that it focuses on genes, which have previously
135
been associated with craniofacial embryogenesis or inherited craniofacial syndromes, rather
136
than screening hundreds of thousands of non-specific markers. This approach aims to
137
increase the chances of finding significant associations between SNPs and visible traits and
138
requires fewer samples for robust association analysis [24, 25].
139
In the current study, 32 anthropometric landmarks were recorded from 3-D facial scans of
140
587 volunteers from general Australian population (Gold Coast, Queensland). Additionally,
141
three direct cranial measurements using a calliper were made and two facial traits (ear lobe
142
and eye lid morphology) were recorded. Both the direct measurements and the Cartesian
143
coordinates of the anthropometric landmarks were used to calculate 92 craniofacial distances.
144
The calculation of 10 principal components based on the craniofacial measurements was
145
performed in order to obtain a more simplified representation of the facial shape. The
146
associations between 104 of the total craniofacial traits and 2,332 genetic markers were
147
tested.
148
This research aims to assist in uncovering the genetic basis of normal craniofacial
149
morphology variation and will enhance our understanding of craniofacial embryogenetics.
150
These findings could be useful in building models to predict facial appearance from a
151
forensic DNA sample where no suspect has been identified, thereby providing valuable
Page 5
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
152
investigative leads. It could also assist in identifying skeletal remains by allowing more
153
accurate facial reconstructions.
154
155
Results
156
157
3D measurements precision study
158
In the last decade 3D scanning systems have been extensively used in anthropometric
159
studies as well as in medical research [18, 20, 64, 65]. The Minolta Vivid V910 3D scanner
160
has been demonstrated to have accuracy to a level of 1.9 ± 0.8 mm [66] and 0.56 ± 0.25 mm
161
[67], making it suitable for the present study since it should provide an accurate
162
representation of facial morphology. However, the allocation of anthropometric facial
163
landmarks can be challenging, especially when tissue palpating is not possible.
164
Reproducibility of the landmark precision was assessed on fifteen 3D facial images through
165
assessment of 85 facial measurements, including linear and angular distances and ratios
166
between the linear distances at two separate times. The period between the analyses varied
167
from one to six months. The mean difference (MD) was calculated as the discrepancy
168
between the first and the second measurement. The measurement error (ME) was calculated
169
as the standard deviation of the MD divided by square root of 2 (ME=SD(MD/√2.
170
In general, the nasal area distances, which involved nasion, pronasale, subnasale and alare
171
landmarks showed greater reproducibility, while the measurements involving paired
172
landmarks, such as gonion and zygion demonstrated higher variance. This result can be
173
explained by easier allocation of nasal area landmarks, compared with gonion and zygion
174
[29]. Overall the median difference (MD) between two measurements for linear distances in
175
15 images ranged between 0.76 mm (ME ±0.27) and 2.80 mm (ME ±0.99); for angular
176
distances between 0.38 mm (ME ±0.96) and 3.75 mm (ME ±0.40) and for facial indices
177
(ratios) between 0.46 mm (ME ±1.08) and 2.98 mm (ME ±1.95) respectively. The lower
178
reproducibility in the angular distances and indices can be explained by a higher number of
179
landmarks (hence variability in allocation of x, y and z coordinates) needed for their
180
calculation (three and four landmarks respectively). Nevertheless, our findings are concordant
181
with the published results, which observed variance of 0.19 mm to 3.49 mm with a ME range
182
of 0.55 mm to 3.34 mm for each landmark [19, 68].
Page 6
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
183
184
Candidate genes search and sequencing data quality control
185
The search for candidate genes and SNPs potentially involved in influencing normal
186
craniofacial morphology variation initially focused on searching for genes involved in normal
187
or abnormal craniofacial variation in humans and model organisms (Supplemental Table S1).
188
As a complementary approach, a search for genetic markers with high Fst values (≥0.45) was
189
implemented, based on the rationale that genes involved in craniofacial morphology
190
regulation are likely to display significant differences in allele frequencies across populations.
191
The first approach has mainly focused on the Mouse Genome Informatics (MGI) database
192
search using the keyword ‘craniofacial mutants’ and additional resources such as Online
193
Mendelian Inheritance in Man (OMIM), GeneCards and AmiGO, using the keywords such as
194
“craniofacial”, “craniofacial mutants”, “craniofacial anomalies”, “craniofacial dimorphism”
195
and “facial morphology” (a detailed list of used resources is summarized in Supplemental
196
Appendix S1). This search revealed a list of 2,891 genotypes and 7,956 annotations. A search
197
of the ‘abnormal facial morphology’ sub-category resulted in 1,492 genotypes and 2,889
198
annotations. The final search of the ‘abnormal nose morphology’ of the previous sub-
199
category revealed 219 genotypes with 310 annotations, representing approximately 150
200
genes.
201
In parallel, a search for high Fst markers, using previously published AIMs and web tools,
202
such as ENGINES, resulted in identification of additional targets, for a total of 1,088 genes
203
and intergenic regions (a detailed list of used resources is summarized in Supplemental
204
Appendix S1).
205
However, manual examination revealed that 592 of these genes showed no apparent link with
206
normal craniofacial development or malformations and were therefore excluded. The
207
remaining 496 regions were further screened for non-synonymous and potentially functional
208
SNPs, as well as SNPs with high population differentiation, which resulted in the shortlist of
209
269 genes and intergenic regions.
210
Subsequent analysis of these 269 genes/regions for functional annotation using the AmiGO
211
Gene Onthology server [57], resulted in 177 candidate genes/regions, possessing 1,319
212
genetic markers involved in various stages of human embryonic development, including:
213
embryonic morphogenesis, sensory organ development, tissue development, pattern
Page 7
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
214
specification process, tissue morphogenesis, ear development, tube morphogenesis,
215
epithelium development, chordate embryonic development and morphogenesis of an
216
epithelium (Supplemental Appendix S1). Notably, the majority of these markers are located
217
in introns and intergenic regions.
218
In terms of molecular function, AmiGO showed that craniofacial candidate markers might be
219
involved in a range of regulatory activities including: protein dimerization activity, chromatin
220
binding, regulatory region DNA binding, sequence-specific DNA binding RNA polymerase
221
II transcription factor activity, sequence-specific distal enhancer binding activity, heparin
222
binding, RNA polymerase II core promoter proximal region sequence-specific DNA binding
223
transcription factor activity involved in positive regulation of transcription, BMP receptor
224
binding and transmembrane receptor protein serine/threonine kinase binding (Supplemental
225
Appendix S1).
226
Subsequent analysis of candidate SNPs for mouse phenotype associations confirmed that
227
orthologous candidate markers were previously detected in mouse models displaying
228
abnormal morphology of the skeleton, head, viscerocranium and facial area, as well as
229
specific malformations of the eye, ear, jaw, palate, limbs, digits and tail (data not shown).
230
In additional to craniofacial candidate SNPs, 522 markers, previously shown to be associated
231
with pigmentation traits, such as eye, skin and hair colour were selected from the relevant
232
literature. These markers were used to validate the results of the genetic association analyses
233
of craniofacial traits.
234
The final candidate marker list was analysed using the GREAT platform to visualize the
235
genomic context of amplicons covering targeted SNPs [69]. The analysis revealed that almost
236
99% of the genomic regions (which may cover multiple markers) are associated with one or
237
two genes with approximately 62% of genomic regions located 0-500 kb downstream of a
238
transcription start site (data not shown).
239
Targeted massively parallel sequencing of the 587 samples resulted in 9,051 genetic markers,
240
with the majority of markers (>5,000) represented by rare polymorphisms of ≤1% minor
241
allele frequency (MAF) (data not shown). The difference between the initial hot-spot SNP
242
panel of candidate markers (n=6,945) and the actual sequencing output (n=9,051) was a result
243
of identification of potentially novel and rare markers in individual DNA samples. Three of
244
the 587 samples, did not produce high quality genotypes because of poor DNA quality or
245
unsuccessful library and template preparation.
Page 8
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
246
The SNPs were filtered by sequencing quality and by MAF. Data quality control was
247
performed by removing markers of low genotype quality (GQ>10) and sequencing depth
248
(DP>10X), which resulted in 8,518 markers (Supplemental Appendix S2). Further filtering of
249
markers using a 2% MAF cut-off resulted in 3,075 markers (Supplemental Appendix S2).
250
The decision to apply a slightly more stringent MAF threshold (2%) was made because of the
251
sample size (n=587) and to reduce potential bias from rare SNPs (1% MAF). Since this may
252
reduce the power of analysis, we analysed and compared both datasets and did not observed
253
any significant difference. Additional filtering based on the HWE threshold of p-value ≥0.01
254
resulted in 2,332 markers. The mean sequencing depth for significantly associated markers in
255
this study was 58 fold (±48.9 SD).
256
257
Genetic association study
258
The association analyses were performed using a linear regression model,
259
incorporating EIGENSTRAT-generated PCA as well as sex and BMI as covariates. The use
260
of covariates in the statistical analysis aimed to reduce the risk of introducing confounding
261
effects, which can result in false positive associations. While sexual dimorphism in the
262
craniofacial morphology is well-known [70], BMI will also likely affect certain craniofacial
263
traits, since the soft facial tissue may change significantly with weight gain or loss. Despite
264
that, this potential confounding factor has to date been disregarded in association studies of
265
normal craniofacial morphology. Age was not considered a significant covariate, given that
266
average age of the subjects in this study was 27 (±8.9 SD). Nevertheless, the potential effect
267
of age as a cofactor was assessed on three craniofacial traits and found to be not significant
268
(data not shown).
269
While the majority of current GWA studies rely on a p-value <5.00E-08 significance
270
threshold, some publications suggest this threshold may be too stringent, especially for
271
complex traits that are regulated by a large number of small effect alleles [75, 76]. In contrast
272
to GWAS, candidate gene studies undertake a more focused genetic strategy, concentrating
273
on a relatively limited number of putative markers. As this study analysed a significantly
274
lower number of SNPs than usual GWA-studies, we could use a higher p-value cut-off since
275
the smaller sample size means the probability of false positive at extremely low p-values is
276
itself lower. Nevertheless, we decided to keep the traditional GWAS p-value significance
277
threshold (<5.00E-08) in order to reduce the possibility of detecting false positive results.
Page 9
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
278
In addition, we subsequently applied a more stringent Bonferroni – corrected threshold in
279
order to minimize the chance of detecting spurious associations. Following the association
280
analysis of 104 craniofacial phenotypes with 2,332 genetic markers, the significance
281
threshold based on the Bonferroni correction with a desired α of 0.05 would be 2.06E-07
282
(=0.05/(2,332*104)).
283
However, it should be emphasized that the Bonferroni correction is widely considered over-
284
conservative, especially in the case of complex phenotypic traits with small individual effects
285
of each allele. Considering that our results confirm the previously published findings, we
286
believe the GWAS p-value threshold is conservative enough to avoid or at least significantly
287
reduce potentially spurious associations. Following this rationale, we report all the variants,
288
which met the unadjusted genome-wide association p-value threshold as suggestive. We
289
believe these findings are useful for the future studies focusing on genetics of normal
290
craniofacial morphology.
291
The results of the association analyses of the craniofacial traits are summarized in Table 1
292
and Supplemental Figs. S1-S16. In general, following the application of a stringent
293
Bonferroni-corrected GWAS threshold (adjusted p-value <1.6E-07), we observed five
294
craniofacial traits being associated with six genomic markers. Specifically, nasal width with
295
rs8035124 (p-value 1.74E-07, Beta=1.366, SE=0.209), cephalic index with rs16830498 (p-
296
value 8.67E-08, Beta=3.005, SE=0.4518), nasal index with rs37369 (p-value 1.43E-07,
297
Beta=4.025, SE=0.6124), transverse nasal prominence angle with rs59037879 (p-value
298
6.07E-09, Beta=4.765, SE=0.6685) and rs10512572 (p-value 1.57E-08, Beta=1.505,
299
SE=0.2171), and principal component (EV=1391.99) with rs37369 (p-value 2.85E-08, Beta=-
300
0.021, SE=0.003079) and rs390345 (p-value 8.55E-08, Beta=-0.0184, SE=0.002768). The
301
polymorphisms: rs16830498, rs59037879 and rs390345 are intronic variants in CACNB4,
302
ZEB1 and FOXN3 respectively; rs37369 is a missense mutation in the AGXT2 gene and
303
rs8035124 and rs10512572 are intergenic variants in 15q12.2 and 17q21.33 chromosomal
304
locations respectively.
305
306
307
Page 10
309
Table 1. Results of genetic association analyses between candidate SNPs and craniofacial traits, including all genomic markers reached the
unadjusted p-value threshold of <5.00E-08.
gene/intergenic region
rs#
chromosomal
location
observed
alleles
MAF
genomic
annotation
UNADJ
BONF
HOLM
BETA
SE
nasal width (al-al)
15q26.1
rs8035124
15:92105708
A/C
3.08E-01
intergenic
1.52E-10
1.74E-07
1.74E-07
1.37E+00
2.09E-01
EYA2
rs58733120
20:45803852
C/G
2.29E-02
intronic
5.37E-10
6.15E-07
6.14E-07
4.98E+00
7.86E-01
RP11-494M8.4
rs1482795
11:7850345
C/T
1.71E-01
intergenic
7.68E-10
8.78E-07
8.77E-07
1.56E+00
2.48E-01
AGXT2
9q22.32 (downstream to
PTCH1)
EYA1
rs37369
5:35037115
C/T
1.77E-01
missense
1.04E-09
1.19E-06
1.19E-06
1.51E+00
2.43E-01
rs57585041
9:98205221
G/T
2.95E-02
intergenic
6.05E-09
6.92E-06
6.90E-06
3.63E+00
6.13E-01
rs79867447
8:72127562
C/T
2.19E-02
intronic
3.92E-08
4.49E-05
4.47E-05
4.89E+00
8.73E-01
17q24.3
rs10512572
17:69512099
intergenic
2.22E-08
2.54E-05
2.54E-05
-9.43E-01
1.66E-01
nasal tip protrusion (sn-prn)
A/G
1.67E-01
cephalic index
CACNB4
rs16830498
2:152814028
C/T
9.06E-02
intronic
7.57E-11
8.67E-08
8.67E-08
3.01E+00
4.52E-01
MYO5A
rs2290332
15:52611451
A/G
2.19E-01
synonymous
5.56E-10
6.37E-07
6.37E-07
1.99E+00
3.15E-01
ZEB1
rs59037879
10:31745993
A/T
2.49E-02
intronic
6.27E-10
7.18E-07
7.17E-07
6.24E+00
9.85E-01
COL11A1
rs4908280
1:103420759
G/T
3.14E-01
intronic
1.66E-09
1.91E-06
1.90E-06
-1.70E+00
2.77E-01
EYA1
rs1481800
8:72131426
A/G
3.62E-01
intronic
2.07E-09
2.37E-06
2.36E-06
1.66E+00
2.72E-01
TEX41
rs10496971
2:145769943
G/T
1.87E-01
intronic
5.32E-09
6.09E-06
6.06E-06
1.99E+00
3.36E-01
PCDH15
rs10825273
10:55968685
C/T
2.82E-01
intronic
9.93E-09
1.14E-05
1.13E-05
1.71E+00
2.94E-01
COL11A1
rs11164649
1:103444679
G/T
3.15E-01
intronic
1.70E-08
1.95E-05
1.94E-05
-1.63E+00
2.85E-01
5q14.3
rs373272
5:84818656
A/G
4.22E-01
intergenic
2.40E-08
2.75E-05
2.73E-05
1.53E+00
2.69E-01
AGXT2
rs37369
5:35037115
C/T
1.77E-01
missense
1.25E-10
1.43E-07
1.43E-07
4.03E+00
6.12E-01
EYA2
rs58733120
20:45803852
C/G
2.29E-02
intronic
9.46E-09
1.08E-05
1.08E-05
1.18E+01
2.03E+00
RP11-408B11.2
rs7311798
12:85808703
C/T
9.71E-02
intergenic
1.77E-08
2.02E-05
2.02E-05
5.00E+00
8.73E-01
11q15.4
rs1482795
11:7850345
C/T
1.71E-01
intergenic
1.83E-08
2.09E-05
2.09E-05
3.66E+00
6.40E-01
EYA1
rs79867447
8:72127562
C/T
2.19E-02
intronic
3.53E-08
4.03E-05
4.02E-05
1.26E+01
2.24E+00
nasal index (al-al/n-sn)
Page 11
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
308
rs79867447
8:72127562
C/T
2.19E-02
intronic
1.10E-09
1.26E-06
1.26E-06
3.75E+00
6.02E-01
EYA2
rs58733120
20:45803852
C/G
2.29E-02
intronic
3.38E-09
3.86E-06
3.86E-06
3.25E+00
5.39E-01
EGFR
rs17335905
7:55131384
C/T
3.30E-02
intronic
4.74E-08
5.42E-05
5.41E-05
2.26E+00
4.07E-01
intronic
2.07E-09
2.36E-06
2.36E-06
-1.15E+01
1.87E+00
nasolabial angle (prn-sn-ls)
SMAD1
rs17020235
4:146418167
A/G
3.59E-02
ZEB1
rs59037879
10:31745993
A/T
2.49E-02
intronic
5.31E-12
6.07E-09
6.07E-09
4.77E+00
6.69E-01
17q24.3
rs10512572
17:69512099
A/G
1.67E-01
intergenic
1.38E-11
1.57E-08
1.57E-08
1.51E+00
2.17E-01
AGXT2
rs37369
5:35037115
C/T
1.77E-01
missense
1.46E-09
1.66E-06
1.66E-06
1.31E+00
2.12E-01
LMNA
rs12076700
1:156055099
C/G
2.28E-01
intronic
1.54E-09
1.75E-06
1.75E-06
1.18E+00
1.92E-01
FAM49A
rs6741412
2:16815759
C/G
3.99E-01
intronic
2.75E-09
3.14E-06
3.13E-06
9.96E-01
1.64E-01
transverse nasal prominence angle (t-l)-prn-(t-r)
TEX41
rs10496971
2:145769943
G/T
1.87E-01
intronic
5.52E-09
6.30E-06
6.27E-06
1.27E+00
2.14E-01
RTTN
rs74884233
18:67813813
A/G
2.59E-02
intronic
1.20E-08
1.37E-05
1.37E-05
3.07E+00
5.30E-01
AC073218.1
rs892458
2:34667749
C/T
4.97E-01
intergenic
1.73E-08
1.98E-05
1.97E-05
9.61E-01
1.68E-01
PAX3
rs2289266
2:223089431
G/T
1.23E-01
intronic
1.95E-08
2.23E-05
2.21E-05
1.58E+00
2.76E-01
LHX8
rs12041465
1:75609049
A/C
2.37E-01
intronic
2.30E-08
2.62E-05
2.60E-05
1.21E+00
2.12E-01
AC073218.1
rs892457
2:34667721
G/A
4.98E-01
intergenic
3.43E-08
3.92E-05
3.89E-05
9.38E-01
1.67E-01
14q22.1 (upstream to BMP4)
rs2357442
14:52607967
A/C
2.03E-01
intergenic
4.40E-08
5.03E-05
4.98E-05
1.11E+00
1.99E-01
PC1 (EV=1391.99)
AGXT2
rs37369
5:35037115
A/G
1.77E-01
missense
2.49E-11
2.85E-08
2.85E-08
-2.10E-02
3.08E-03
FOXN3
rs390345
14:89976534
A/G
2.47E-01
intronic
7.46E-11
8.55E-08
8.54E-08
-1.84E-02
2.77E-03
FAM49A
rs6741412
2:16815759
G/A
3.99E-01
intronic
4.67E-10
5.35E-07
5.34E-07
-1.52E-02
2.40E-03
17q24.3
rs10512572
17:69512099
G/A
1.67E-01
intergenic
4.99E-10
5.71E-07
5.70E-07
-2.01E-02
3.17E-03
EYA1
rs79867447
8:72127562
C/T
2.19E-02
intronic
7.46E-10
8.54E-07
8.51E-07
-6.45E-02
1.03E-02
LMNA
rs12076700
1:156055099
C/G
2.28E-01
intronic
7.87E-10
9.01E-07
8.97E-07
-1.73E-02
2.76E-03
PCDH15
rs10825273
10:55968685
C/T
2.82E-01
intronic
1.04E-09
1.19E-06
1.19E-06
-1.68E-02
2.70E-03
14q22.1 (upstream to BMP4)
rs942316
14:54440983
A/C
1.21E-01
intergenic
2.66E-09
3.04E-06
3.02E-06
-2.36E-02
3.89E-03
TEX41
rs10496971
2:145769943
T/G
1.87E-01
intronic
3.71E-09
4.25E-06
4.22E-06
-1.84E-02
3.06E-03
Page 12
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
nose-face width index (al-al/zy-zy)
EYA1
rs7844723
8:122908503
C/T
4.37E-01
intergenic
8.11E-09
9.29E-06
9.21E-06
1.41E-02
2.41E-03
EYA2
rs58733120
20:45803852
G/C
2.29E-02
intronic
2.19E-08
2.51E-05
2.49E-05
-5.65E-02
9.94E-03
FAM49A
rs11096686
2:16815892
T/C
2.50E-01
intronic
2.25E-08
2.57E-05
2.55E-05
-1.65E-02
2.89E-03
EYA1
rs73684719
8:72131359
G/A
2.59E-02
intronic
2.62E-08
3.00E-05
2.96E-05
-4.94E-02
8.75E-03
XXYLT1
rs950257
3:194847650
T/A
3.87E-01
intronic
3.94E-08
4.51E-05
4.46E-05
-1.38E-02
2.48E-03
310
311
Highlighted with bold: genomic markers that reached Bonferroni corrected threshold (1.6E-07); Highlighted with blue colour: linear distances; Highlighted
312
with red colour: craniofacial indices; Highlighted with green colour: angular distances; Highlighted with violet colour: principal component; gene/intergenic
313
region: gene name/locus; rs#: reference SNP ID number; chromosomal location: chromosomal location of the marker based on the GRCh37/hg19; observed
314
alleles: common alleles observed in human genome based on dbSNP build 147; MAF: minor allele frequency; genomic annotation: genomic location of the
315
marker; UNADJ: Unadjusted p-values. BONF: Bonferroni single-step adjusted. HOLM: Holm (1979) step-down adjusted; BETA: minor allele effect size;
316
SE: standard error.
317
318
319
Page 13
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
RP11-785H20.1
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
320
However, given the over-conservative nature of the Bonferroni correction and also assuming
321
that polygenic traits are likely to be dominated by numerous alleles with small causal effect
322
we also report suggestive associations reaching the unadjusted 5.00E-08 p-value threshold
323
(Table 1). Generally, two linear distances (nasal width and nasal tip protrusion), two angular
324
distances (nasolabial angle and transverse nasal prominence angle), three indices (cephalic
325
index, nasal index and nose-face width index) and one principal component revealed
326
significant associations with 34 SNPs in 28 genes and intergenic regions (Table 1).
327
These factors can be arbitrarily divided into three main categories based on their cellular
328
function: 1) genes with known roles in the craniofacial morphogenesis and/or mutated in
329
various hereditary syndromes displaying craniofacial abnormalities; 2) genes or pseudo-genes
330
without known function in the craniofacial morphology regulation or previously
331
uncharacterized genes; and 3) non-protein coding genes, such as lncRNA class genes. There
332
are also a number of significant variants that are located in the intergenic regions, with or
333
without proximity to open reading frames (ORFs).
334
The majority of associated markers (n=21) are located in 17 protein-coding genes and
335
pseudo-genes such as AGXT2, CACNB4, COL11A1, EGFR, EYA1, EYA2, FAM49A, FOXN3,
336
LHX8, LMNA, MYO5A, PAX3, PCDH15, RTTN, SMAD1, XXYLT1 and ZEB1.
337
Five variants are present in RNA-coding (lncRNA) genes, which include AC073218.1, RP11-
338
494M8.4, RP11-408B11.2, RP11-785H20.1 and TEX41.
339
The rest of the markers (n=6) are found in the intergenic regions, near the following genes
340
and pseudogenes: BMP4, HAS2-AS1, LOC124685, LOC100131241, MRPS36P3, PTCH1,
341
SLC25A5P2, SV2B and TRNAY16P.
342
Analysis of the functional annotation of significant markers revealed that one SNP represent
343
missense mutation (rs37369), one SNP is a synonymous transversion (rs2290332), 21
344
markers are located in intronic sequences and 11 markers are located in intergenic regions
345
(Table 1). The majority of significantly associated SNPs (n=27) are found in the regulatory
346
elements of the genome, such as in transcription factor (TF) binding sites, and represent
347
potentially functional SNPs (pfSNPs). These variants may be involved in “fine tuning” of the
348
normal craniofacial phenotype as part of the enhancer/silencer mechanisms, as has been
349
recently suggested [77].
350
The nasal area measurements, using either “n”, “prn”, “sn” or “al” landmarks, produced the
351
majority of the total number of significant associations (6 out of 8). These measurements
Page 14
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
352
include nasal width (al-al), nasal tip protrusion (sn-prn), nasolabial angle (prn-sn-ls),
353
transverse nasal prominence angle (t_l-prn-t_r), nasal index (al-al/n-sn), and nose face width
354
index (al-al/zy-zy). The apparent overrepresentation of associations with the nasal area may
355
be a result of the easier allocation and consequent superior reproducibility of the nasal area
356
landmark measurements on 3D images. It may also be the result of specific selection of
357
candidate genes from the JAX mice database resource, which focused on mutants that
358
displayed various nasal area abnormalities.
359
The analysis of direct cranial measurements and their relative indices revealed significant
360
associations only the Cephalic index (CI) with 9 SNPs.
361
The association analysis of the principal components (PC) representing all the craniofacial
362
measurements, revealed one principal component (explaining 73.3% of all the craniofacial
363
phenotypes) that was associated with 14 genetic markers (Table 1).
364
In contrast to most other craniofacial association studies that focused on a specific
365
homogeneous population group (mostly Europeans), this study included samples from several
366
population groups, which enabled investigation of the genetic factors influencing normal
367
craniofacial morphology in different ethnicities [71]. Self-reported ancestry however, cannot
368
be considered fully reliable, as demonstrated previously [72, 73]. In order to address this
369
issue we assessed the self-reported ancestry using STRUCTURE with 186 SNPs removed
370
due to long-range disequilibrium [49]. Following the rationale that the best ancestry estimates
371
are obtained using a large number of random markers [74], we used all the available markers
372
(after MAF filtering) in STRUCTURE analysis. The STRUCTURE analysis resulted in
373
clusters of 367 Europeans, 51 East Asians, 43 South Asians and 16 Africans, with 107
374
samples designated as admixed ancestry (Fig. 1). Of the samples tested with STRUCTURE,
375
459 (89%) were assigned the same ancestry cluster (sole or mixed origin) as the self-reported
376
information. Of the remaining 57 individuals, 39 were estimated as ‘admixture’ (based on up
377
to 20% admixture threshold) and 18 were assigned a single ancestry, different to the self-
378
reported ancestry (Fig. 1).
379
The risk of detecting false positive results because of population stratification was carefully
380
assessed and further reduced by applying an EIGENSTRAT correction. Specifically,
381
EIGENSTRAT’s smartpca.perl was used to perform PCA-clustering in comparison to
382
reference populations from HapMap reference clusters. The Q-Q plots of the associated traits
383
showed the expected distribution of data after applied correction (Supplemental Figs. S1-S8).
Page 15
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
384
We did not perform allele imputations on this dataset because it includes individuals from
385
heterogenous ancestral backgrounds, with 107 subjects classified as 'admixture', based on the
386
applied threshold of 20%. Imputation using homogenous reference populations would have
387
introduced unnecessary bias with wrongly imputed alleles in subsequent analysis steps.
388
389
Association analyses of the non-craniofacial traits
390
In our attempt to identify genetic markers influencing normal variation in craniofacial traits,
391
we incorporated 522 markers previously associated with human pigmentation traits, such as
392
eye, skin and hair colour. These markers were included to validate the statistical methods
393
used for the craniofacial traits association study. The association analyses of the pigmentation
394
traits, which were based on the HWE non-filtered data, did indeed confirm previously
395
published findings, as detailed in Table S2. It should be noted however, that these results may
396
not necessarily confirm the validity of the craniofacial markers associations.
397
The application of the Hardy-Weinberg equilibrium (HWE) threshold resulted in filtering
398
25% of the total number of SNPs. These markers included almost all the SNPs, previously
399
associated with pigmentation traits, such as rs12913832, rs1129038, rs8039195 and
400
rs16891982. This is not surprising, since population-related markers are likely not being in
401
HWE ‘a priori’. Another explanation for this observation is potential bias from partially
402
uncorrected heterogeneous ancestry, since the ancestry correction algorithm can only
403
minimize, rather than completely remove spurious associations [52]. In fact, the association
404
analyses of the HWE non-filtered genotyping data with pigmentation traits (eye, skin and hair
405
colour), demonstrated highly significant associations, concordant with the literature (Table
406
S2).
407
408
Craniofacial gene and SNP annotations
409
The following section summarizes the genetic association results, providing brief
410
annotation of the significantly associated genes and SNPs. Functional annotations, such as
411
predicted molecular function, link to a biological process and a protein class of the 23
412
protein-coding genes and pseudo-genes (AGXT2, BMP4, CACNB4, COL11A1, EGFR, EYA1,
413
EYA2, FAM49A, FOXN3, LHX8, LMNA, MYO5A, PAX3, PCDH15, RTTN, SMAD1, XXYLT1
Page 16
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
414
and ZEB1) have been visualised using the PANTHER resource [78] and summarized in
415
supplemental materials (Supplemental Figs. S17-S19).
416
Significantly associated genes with previously demonstrated role in
417
craniofacial morphogenesis and/or mutated in hereditary syndromes
418
displaying craniofacial abnormalities
419
A potentially functional SNP rs2289266 in the intron of the Paired Box 3 gene (PAX3) was
420
associated with the transverse nasal prominence angle (p-value 1.95E-08). This gene is a
421
member of the paired box (PAX) family of transcription factors, which play critical roles
422
during foetal development. The PAX3 protein regulates cell proliferation, migration and
423
apoptosis. Mutations in PAX3 are associated with Waardenburg syndrome (OMIM: 193500),
424
which is characterized by a prominent and broad nasal root, a round or square nose tip,
425
hypoplastic alae, increased lower facial height and other craniofacial abnormalities.
426
Notably, three other SNPs in this gene, rs974448, rs7559271 and rs1978860, were previously
427
associated with normal variability of the nasion position [22] and the distance between the
428
eyeballs and the nasion [20]. None of these SNPs were included in this study, as a result of
429
primer design failure. No LD between rs2289266 and any of the previously associated
430
markers in the PAX3 gene was detected. Nevertheless, the association of another variant in
431
the PAX3 gene can be considered an independent confirmation of this gene’s involvement in
432
regulation of normal craniofacial morphology.
433
SNPs rs4908280 and rs11164649 which are located in the regulatory element of the Collagen
434
gene (COL11A1) intronic sequence, were associated with the cephalic index (p-values 1.66E-
435
09 and 1.70E-08 respectively). COL11A1 encodes one of the two alpha chains of type XI
436
fibrillar collagen and is known to have multiple transcripts as a result of alternative splicing.
437
The secreted protein is hypothesised to play an important role in fibrillogenesis by controlling
438
lateral growth of collagen II fibrils.
439
Notably, the same variant (rs11164649) was recently linked to normal-range effects in
440
various craniofacial traits, specifically eyes, orbits, nose tip, lips, philtrum and lateral parts of
441
the mandible, although the measurements of the cephalic index were not performed in this
442
study [93]. Our findings should be considered as independent confirmation of COL11A2 gene
443
and its specific polymorphism rs11164649 involvement in shaping the normal craniofacial
444
morphology.
Page 17
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
445
According to the MGI database, transgenic mice with shortened COL11A2 mRNA (the
446
second alpha chain of type XI fibrillar collagen) display abnormal facial phenotypes,
447
including a triangular face and shorter and dimpled nasal bones [30]. Interestingly, COL11A1
448
and other Collagen family genes were found to be mutated in Stickler (OMIM: 604841) and
449
Marshall Syndromes (OMIM: 154780). These two inherited disorders display very similar
450
phenotypes and each is characterized by a distinctive facial appearance, with flat midface,
451
very small jaw, cleft lip/palate, large eyes, short upturned nose, eye abnormalities, round face
452
and short stature. However, the facial features of Stickler syndrome are less severe and
453
include a flat face with depressed nasal bridge and cheekbones, caused by underdeveloped
454
bones in the middle of the face. Another member of the collagen family, COL17A1, was
455
recently associated with the distance between the eyeballs and the nasion [23]. Our finding of
456
genetic associations of additional members of the Collagen family provides further evidence
457
of the importance of polymorphisms in these genes in determining the normal variety of
458
specific craniofacial features.
459
Intergenic SNP rs942316, which is located upstream to the Bone Morphogenetic Protein
460
(upstream to BMP4) gene, was strongly associated with the PC1phenotype (p-value 2.66E-
461
09). The BMP4 gene is a transforming growth factor, belonging to the beta superfamily,
462
which includes large families of growth and differentiation factors. This gene plays an
463
important role in the onset of endochondral bone formation in humans, including induction of
464
cartilage and bone formation and specifically tooth development and limb formation. Gene
465
onthology annotations related to this gene include heparin binding and cytokine activity.
466
BMP4 mutations have been associated with a variety of bone diseases, including orofacial
467
cleft 11 (OMIM: 600625), Fibrodysplasia Ossificans (OMIM: 135100) and microphthalmia
468
syndromic 6 (OMIM: 607932).
469
SNP rs2290332 represents a synonymous variant in the Myosin VA (Heavy Chain 12,
470
Myoxin) gene (MYO5A). This variant was associated with the cephalic index (p-value 5.56E-
471
10).
472
MYO5A is one of three myosin V heavy-chain genes, belonging to the myosin gene
473
superfamily. Myosin V is a class of actin-based motor proteins involved in cytoplasmic
474
vesicle transport and anchorage, spindle-pole alignment and mRNA translocation. It mediates
475
the transport of vesicles to the plasma membrane, including melanosome transport. Mutations
476
in this gene were associated with a number of neuroectodermal diseases, such as Griscelli
477
syndrome. Additional mutations in this gene were associated with a rare inherited condition
Page 18
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
478
Piebaldism (OMIM:172800). The symptoms of Piebaldism include partial albinism and
479
anomalies of the mouth area development, such as lips and philtrum abnormalities. Despite
480
being a “silent” mutation, rs2290332 is located in the POLR2A TF binding site and may
481
therefore affect various processes such as transcription, translation, splicing and mRNA
482
transport, as has been shown in other studies [79].
483
Variant rs12041465, which is located in the intron of LIM Homeobox 8 (LHX8) was
484
associated with transverse nasal prominence angle (p-value 2.30E-08). LHX8 is a
485
transcription factor and a member of the LIM homeobox family of proteins, which are
486
involved in patterning and differentiation of various tissue types. Mutations in this gene were
487
associated with clefts of the secondary palate in mouse model [80, 81].
488
Three intronic SNPs in the Eyes Absent Homolog 1 (EYA1) gene were associated with
489
several craniofacial traits. The variant rs79867447 was associated with the nose width (p-
490
value 3.92E-08), nasal index (p-value 3.53E-08), nose-face width index (p-value 1.10E-09)
491
and PC1 (p-value 7.46E-10). The variant rs1481800 was associated with the cephalic index
492
(p-value 2.07E-09). The variant rs73684719 was found in association with PC1 (p-value
493
2.62E-08). All three variants belong to potentially regulatory elements of the genome and are
494
likely to affect TF binding sites. No linkage disequilibrium has been detected between these
495
markers.
496
The EYA1 encoded protein functions as histone phosphatase, regulating transcription during
497
organogenesis in kidney and various craniofacial features such as branchial arches, eye and
498
ear. eya1 mutated mice display various craniofacial anomalies of the inner ear, mandible,
499
maxilla and reduced skull [30]. Mutations in the human ortholog have been associated with
500
several craniofacial conditions such as otofaciocervical syndrome (OMIM:166780), Weyers
501
acrofacial dysostosis (OMIM:193530) and branchiootic syndrome (OMIM:608389).
502
Intronic SNP rs58733120 was associated with the nose width (p-value 5.37E-10), nasal index
503
(p-value 9.46E-09), nose-face width index (p-value 3.38E-09) and PC1 (p-value 2.19E-08)
504
phenotypes. This variant is located in the regulatory element of the EYA2 gene, which
505
belongs to the same eyes absent protein family as EYA1 and plays a similar role in the
506
embryonic development. An orthologue eya2 gene encodes a transcriptional activator in mice
507
and may play a role in eye development. Both EYA1 and EYA2 genes were shown to be
508
expressed in the ninth week of human embryonic development [82]. None of the human
509
craniofacial disorders were associated with EYA2 gene to date.
Page 19
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
510
SNP rs12076700 in the intron of the Lamin A gene (LMNA) was associated with the
511
transverse nasal prominence angle (1.54E-09) and PC1 (p-value 7.87E-10).
512
LMNA, together with other Lamin proteins, is a component of a fibrous layer on the
513
nucleoplasmic side of the inner nuclear membrane, which provides a framework for the
514
nuclear envelope and also interacts with chromatin. LMNA encoded protein acts to disrupt
515
mitosis and induces DNA damage in vascular smooth muscle cells, leading to mitotic failure,
516
genomic instability, and premature senescence of the cell. This gene has been found mutated
517
in Mandibuloacral Dysplasia which is characterized by various skeletal and craniofacial
518
abnormalities, including delayed closure of the cranial sutures and undersized jaw [83].
519
Variant rs74884233 was associated with the transverse nasal prominence angle (p-value
520
1.20E-08). This variant is located in the intron of the Rotatin gene (RTTN). RTTN gene is
521
involved in the maintenance of normal ciliary structure, which in turn effects the
522
developmental process of left-right organ specification, axial rotation, and perhaps notochord
523
development.
524
SNP rs17020235 was associated with the nasolabial angle (p-value 2.07E-09). This
525
potentially functional variant is located in the intron of the SMAD Family Member 1 gene
526
(SMAD1). SMAD1 is a transcriptional modulator activated by BMP (bone morphogenetic
527
proteins) type 1 receptor kinase, which is involved in a range of biological activities
528
including cell growth, apoptosis, morphogenesis, development and immune responses.
529
SMAD1 mutant mice display anterior truncation of the head with only one brachial arch
530
present. In human, SMAD1 mutations (together with RUNX2), are associated with the
531
Cleidocranial Dysplasia (OMIM:119600), which is a Craniosynostosis-type disorder
532
affecting cranial bones, palate and other tissues.
533
SNP rs950257 was associated with the PC1 trait (p-value 3.94E-08). This intronic variant is
534
located in the XXYLT1 gene, which codes for Xyloside Xylosyltransferase 1. This protein is
535
an Alpha-1,3-xylosyltransferase, which elongates the O-linked xylose-glucose disaccharide
536
attached to EGF-like repeats in the extracellular domain of Notch proteins signalling
537
network. Notch proteins are the key regulators of embryonic development, which
538
demonstrate a highly conserved sequence in various species. Interestingly, mutations in
539
Notch proteins are associated with Hajdu–Cheney syndrome (OMIM:10250) and Alagille
540
syndrome (OMIM:118450). The main phenotypic symptoms of these conditions include
Page 20
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
541
various malformations of the craniofacial tissues, including broad, prominent forehead, deep-
542
set eyes and a small pointed chin.
543
SNP rs17335905 was associated with the nose-face width index (p-value 4.74E-08). This
544
potentially functional variant is located in the intron of the EGFR gene, which encodes the
545
Epidermal Growth Factor Receptor. EGFR is a cell surface protein that binds to epidermal
546
growth factor (EGF). Binding of the protein to a ligand induces activation of several
547
signalling cascades and leads to cell proliferation, cytoskeletal rearrangement and anti-
548
apoptosis. Mouse carrying mutations in EGFR, express short mandible and cleft palate.
549
550
Significantly associated SNPs, located in genes or pseudo-genes that were
551
not linked to craniofacial morphology regulation or genes with unknown
552
function
553
Intronic variant rs59037879 in the Zinc Finger E-Box Binding Homeobox 1 (ZEB1) was
554
found associated with cephalic index (p-value 6.27E-10), and transverse nasal prominence
555
angle (p-value 5.31E-12). This gene encodes a zinc finger transcription factor, which is a
556
transcriptional repressor. It regulates expression of different genes, such as interleukin-2 (IL-
557
2) gene, ATPase transporting polypeptide (ATP1A1) gene and E-cadherin (CDH1) promoter
558
in various cell types and also represses stemness-inhibiting microRNA. Mutations in this
559
gene were previously associated with Corneal Dystrophy and various types of cancer.
560
A missense mutation rs37369 in the Alanine--Glyoxylate Aminotransferase 2 gene (AGXT2)
561
was associated with nose width (p-value 1.04E-09), nasal index: (p-value 1.25E-10),
562
transverse nasal prominence angle (p-value 1.46E-09) and PC1 (p-value 2.49E-11). This
563
protein plays an important role in regulating blood pressure in the kidney through
564
metabolizing asymmetric dimethylarginine (ADMA), which is an inhibitor of nitric-oxide
565
(NO) synthase.
566
An intronic SNPs rs16830498, located in the regulatory element of the Calcium Channel
567
Voltage-Dependent Beta 4 Subunit (CACNB4) gene intron, were significantly associated with
568
cephalic index (p-value 7.57E-11).
569
The beta subunit of voltage-dependent calcium channels may increase peak calcium current
570
by shifting the voltage dependencies of activation and inactivation, modulating G protein
Page 21
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
571
inhibition and controlling the alpha-1 subunit membrane targeting. CACNB4 may be
572
expressed in different isoforms through alternative splicing. Certain mutations in this gene
573
have been associated with various forms of epilepsy, although no association with normal or
574
abnormal craniofacial variation has been previously reported.
575
Potentially functional intronic SNP rs10825273 located in the regulatory elements of the
576
Protocadherin-Related 15 (PCDH15) gene, was found in association with cephalic index (p-
577
value 9.93E-09) and PC1 (p-value 1.04E-09). PCDH15 is a member of the cadherin
578
superfamily, which encodes an integral membrane protein that mediates calcium-dependent
579
cell-cell adhesion and is known to have numerous alternative splicing variants. It plays an
580
essential role in the maintenance of normal retinal and cochlear function. Mutations in this
581
gene result in hearing loss and are associated with Usher Syndrome Type IIA (OMIM:
582
276901).
583
Two intronic variants in the Family With Sequence Similarity 49 Member A gene (FAM49A)
584
were associated with multiple craniofacial traits. rs6741412 was found in association with the
585
transverse nasal prominence angle (p-value 2.75E-09) and PC1 (p-value 4.67E-10).
586
rs11096686 was associated with PC1 (p-value 2.25E-08). The FAM49A protein is known to
587
interact with hundreds of miRNA molecules during pre-implantation of the mouse embryo
588
and also expressed in the developing chick wing, but no information on its specific function
589
or disease association have been identified.
590
SNP rs390345, located in the intronic regulatory sequence of the Forkhead Box N3 gene
591
(FOXN3), was associated with the PC1 (p-value 7.46E-11). FOXN3 encodes multiple splicing
592
variants and acts as a transcriptional repressor. It is proposed to be involved in DNA damage-
593
inducible cell cycle arrests at G1 and G2. There are no previous reports on FOXN3
594
association with either normal craniofacial development or pathological conditions.
595
596
Significantly associated SNPs located in the non-protein coding genes, such
597
as lncRNA class genes
598
Intronic SNP rs10496971 in the TEX41 (Testis Expressed 41) gene produced significant
599
associations with transverse nasal prominence angle (p-value 5.52E-09), cephalic index (p-
600
value 5.315E-09) and PC1 (p-value 3.71E-09).
Page 22
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
601
TEX41 is a long intergenic non-protein coding RNA (lncRNA) class gene, which is located
602
on chromosome 2 and has 43 transcript variants as a result of alternative splicing. lncRNAs
603
are known as regulators of diverse cellular processes. However, the function of this gene
604
remains unknown. Despite its name, this gene is expressed in a variety of tissues, with the
605
highest demonstrated levels in kidney. Its potential involvement in craniofacial genetics, and
606
specifically in influencing normal facial variation, has not been reported previously. Notably,
607
the rs10496971 variant is located in the regulatory element of the genome (as well as 49 other
608
associated SNPs) and may influence normal craniofacial morphology by affecting either
609
enhancer or silencer sequences or transcriptional factor (TF) binding sites [77].
610
The SNP rs1482795, located in the RNA gene RP11-494M8.4, was associated with the nose
611
width (p-value 7.68E-10) and nasal index (p-value 1.83E-08) measurements.
612
Both SNPs rs892457 and rs892458 located in the non-protein coding lncRNA gene
613
AC073218.1, were associated with the transverse nasal prominence angle (p-value 3.43E-08)
614
and (p-value 1.73E-08), respectively.
615
SNP rs7311798, located in the lncRNA gene RP11-408B11.2 was associated with the nasal
616
index (p-value 1.77E-08).
617
SNP rs7844723 in the RP11-785H20.1 (lncRNA gene) was associated with the PC1 (p-value
618
8.11E-09) phenotype.
619
SNP rs2357442 was associated with the transverse nasal prominence angle (p-value 4.40E-
620
08). This variant is located in the Long Interspersed Nuclear Element 1 (LINE-1)
621
retrotransposon sequence, which in turn shows homology with uncategorized mRNA
622
KC832805 on the Y-chromosome.
623
LINE-1 elements comprise approximately 21% of the human genome, and have been shown
624
to modulate expression and produce novel splice isoforms of transcripts from genes that span
625
or neighbour the LINE-1 insertion site. In addition, rs2357442 is located close to three
626
pseudo-genes with unknown function: SLC25A5P2, LOC100130842 and RP11-1033H12.1,
627
while the last two represent RNA-coding lncRNA genes.
628
Significantly associated SNPs located in the intergenic regions
629
SNP rs10512572, located between Serpine1 MRNA Binding Protein 1 pseudogene
630
(LOC100131241) and MyosinLight Chain 6 Alkali Smooth Muscle and Non-Muscle
631
pseudogene (LOC124685), was associated with nasal tip protrusion (p-value 2.22E-08),
Page 23
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
632
transverse nasal prominence angle (p-value 1.38E-11) and PC1 (p-value 4.99E-10). While
633
pseudogenes in general are non-protein coding, their sequences can be functional and play
634
important roles in different biological processes [85]. It should be noted that some genes may
635
be incorrectly defined as pseudogenes, based solely on their sequence computational analysis
636
[86]. The function of these two pseudogene sequences is unknown.
637
SNP rs8035124 was significantly associated with the nose width (p-value 1.52E-10). This
638
variant is located between the Synaptic Vesicle Glycoprotein 2B (SV2B) and Transfer RNA
639
Tyrosine 16 (Anticodon GUA) Pseudogene (TRNAY16P) genes. The SV2B is a protein
640
coding gene, which plays a role in the control of regulated secretion in neural and endocrine
641
cells. The TRNAY16P is a pseudogene with unknown function.
642
Additional SNP rs373272 was associated with cephalic index (p-value 2.40E-08). However,
643
no genes were identified within 50 kb window of its chromosomal location.
644
645
Discussion
646
This study focused on the identification of genetic markers in a set of candidate genes
647
associated with various craniofacial traits, representing the most comprehensive scan for
648
genetic markers involved in normal craniofacial development performed to date. We
649
identified 8 craniofacial significantly associated (unadjusted p-value < 5.00E-08) with 34
650
genomic variants in 28 genes and intergenic regions. Following the application of Bonferroni
651
correction (adjusted p-value threshold of 1.6E-07), associations were observed between 5
652
craniofacial traits (nasal width, cephalic index, nasal index, transverse nasal prominence
653
angle and principal component) and 6 SNPs (rs8035124, rs16830498, rs37369, rs59037879,
654
rs10512572 and rs390345) located in 6 genes and intergenic regions (15q26.1, 17q24.3,
655
CACNB4, AGXT2, ZEB1 and FOXN3 respectively). We report all the significant markers that
656
met the less stringent GWAS threshold (p-value<5.00E-08), as Bonferroni correction is
657
generally considered over-conservative, especially when analysing complex traits such as
658
craniofacial morphology, which is likely to be influenced by a large number of alleles with
659
relatively small individual effect, similar to height [89, 90].
660
The association of the PAX3 gene and the COL11A1 gene with transverse nasal prominence
661
angle and cephalic index respectively, confirms previous findings [11, 22, 23, 91]. In fact, an
662
intronic SNP rs11164649 that was associated with cephalic index in the current study, was
Page 24
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
663
recently associated with normal-range effects in various craniofacial traits and used for their
664
prediction [91], while the other variants in COL11A1 (rs4908280) and in PAX3 (rs2289266)
665
have not been reported previously. The rest of the identified associations are also novel.
666
These include 21 significantly associated markers in protein-coding genes and pseudo-genes,
667
such as AGXT2, CACNB4, EGFR, EYA1, EYA2, FAM49A, FOXN3, LHX8, LMNA, MYO5A,
668
PCDH15, RTTN, SMAD1, TEX41, XXYLT1 and ZEB1. Additional 7 significantly–associated
669
SNPs are found in intergenic regions adjacent to several loci, such as BMP4, LOC124685,
670
LOC100131241 and PTCH1. Some of these genes were previously linked to craniofacial
671
embryogenesis, while others represent novel associations.
672
Six genetic variants were found in lncRNA genes, which have not been previously linked to
673
craniofacial morphogenesis before. These findings may suggest there may be a yet
674
unexplored level of epigenetic regulation affecting craniofacial morphology. lncRNAs are a
675
recently discovered class of factors, whose expression is thought to be important for the
676
regulation of gene expression through several different mechanisms involving competition
677
with transcription by recruitment of specific epigenetic factors to promoter regions, as well as
678
indirectly affecting gene expression by interacting with miRNA and other cellular factors
679
[92]. The comprehensive role of epigenetic regulation in general, and in craniofacial
680
embryonic development in particular, is poorly understood. There is a limited number of
681
recent studies revealing thousands of enhancer sequences, predicted to be active in the
682
developing craniofacial complex in mice [77, 93] and potentially in humans. Both the
683
epistatic and epigenetic interactions may represent a more complex level of craniofacial
684
morphology regulation and require further investigation.
685
Even though a relatively high number of phenotypes were studied (92 linear and angular
686
measurements and indices), this may still represent an oversimplification of the complexity of
687
the human face. Despite the importance of the association between specific 3D measurements
688
and SNPs demonstrated in this study, the association of facial shapes, represented by the
689
principal components should better represent the face. Given that embryonic developmental
690
processes such as cell proliferation, polarity orientation and migration occur in a 3D
691
environment, principal components that in essence denote specific facial shapes, may provide
692
a more accurate representation of these processes. However, only one of the 10 principle
693
components showed significant associations at the GWAS threshold level. While the
694
explanation of this observation is unclear, it is consistent with other similar studies [22, 23].
695
The specific anthropometric measurements on the other hand, produced numerous significant
Page 25
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
696
associations, identifying many genes and intergenic regions that appear to play important
697
roles in the development of normal human facial appearance. The major limitation of this
698
study is the replication of these results that has not been performed yet due to time and
699
budget constraints. However, the confirmation of the two previously associated genes (PAX3
700
and COL11A1) supports the validity of our findings.
701
Given the high complexity of the face, as well as the composite nature of the genetic
702
regulation that affects its development, alternative comprehensive approaches of capturing
703
facial morphology would be beneficial. A number of such methods has recently revealed
704
additional genes with specific polymorphisms associated with the development of
705
craniofacial traits within the normal variation range [91, 94]. Further studies may involve the
706
use of these or alternative methods to capture the majority of variation in craniofacial traits.
707
Craniofacial phenotypes, together with additional external visible traits such as sex, age and
708
BMI and ancestry, could be treated as a “vector”, which could then be used to predict
709
appearance [95].
710
A recent attempt to predict facial appearance was performed using only 24 SNPs [96]. This
711
approach has promise, although it is largely based on reconstruction of a ‘facial composite
712
image’ through prediction of ancestry, sex, pigmentation and human perception of faces. This
713
approach is reasonable, but it does not negate the use of association studies looking at
714
specific craniofacial traits. Genetic association studies of a large scope of individual
715
anthropometric measurements are essential to provide information on specific genes and their
716
polymorphisms, which affect these traits and may therefore be useful in predicting the size
717
and the shape of specific facial features.
718
Additional association studies on large sample sizes, incorporating dense SNP panels or
719
whole genome sequencing approaches, in conjunction with either a comprehensive set of
720
anthropometrical measurements or morphologically adequate representation of the
721
craniofacial characteristics would be a valuable adjunct to the promising results obtained in
722
this study. These studies will not only improve our understanding of the genetic factors
723
regulating craniofacial morphology, but will also enable a better prediction of the visual
724
appearance of a person from DNA.
725
Page 26
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
726
Methods
727
728
Sample collection and ethics statement
729
A total of 623 unrelated individuals, mostly Bond University (Gold Coast, Australia)
730
students, of Australian ancestry were recruited. The participants provided their written
731
informed consent to participate in this study, which was approved by the Bond University
732
Ethics committee (RO-510). To minimize any age-related influences on facial morphology
733
the samples were largely collected from volunteers aged between 18 and 40. The mean age of
734
the volunteers was 26.6 (SD ± 8.9). Following the exclusion of the individuals who had
735
experienced severe facial injury and/or undergone facial surgery (e.g. nose or chin plastics)
736
587 samples remained for the further step of DNA sequencing.
737
Each participant donated four buccal swabs (Isohelix, Cell Projects, Kent, UK). 3-Dimentional
738
(3-D) facial scans and three direct cranial measurements were obtained as described below.
739
Samples with low DNA quantity or low quality facial scans were eliminated leaving 587
740
DNA samples for subsequent genotyping.
741
Additional phenotypic trait information such as height, weight, age, sex, self-reported
742
ancestry (based on the grandparents from both sides), eye lid (single or double), ear lobe
743
(attached or detached), hair texture (straight, wavy, curly or very curly), freckling (none,
744
light, medium or extensive), moles (none, few or many), as well as eye skin, and hair
745
pigmentation was collected by a single examiner in order to reduce potential variation. The
746
pigmentation traits were arbitrary assigned according to previously published colour charts
747
[26-28].
748
749
3D images collection and analysis
750
Craniofacial scans were obtained using the Vivid 910 3-D digitiser (Konica Minolta,
751
Australia) equipped with a medium range lens with a focal length of 14.5 mm. The scanner
752
output images were of 640 x 480 pixels resolution for 3D and RGB data. Two daylight
753
fluorescent sources (3400K/5400K colour temperature) were mounted at approximately 1.2
754
meters from the subject’s head to produce ambient light conditions.
Page 27
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
755
The scanner was mounted approximately one meter from the volunteer’s head. Each
756
volunteer remained in an upright seated position and kept a neutral facial expression during
757
the scan. Subjects with long hair pulled their hair behind the ears or were asked to wear a hair
758
net. Glasses and earrings were removed.
759
Each volunteer was scanned from a distance of approximately one meter from three different
760
angles (front and two sides). The final merged 3D image was produced by semi-automatically
761
aligning the three scans and manually cropping non-overlapping or superfluous data such as
762
the neck area and hair using Polygone® software (Qubic, Australia). The complete
763
coordinates of each merged 3D image were then saved in a ‘vivid’ file format (vvd) and
764
exported to Geomagic® software (Qubic, Australia) for subsequent image processing.
765
Based on the anthropometrical literature [29] 32 anthropometrical landmarks were manually
766
identified on each 3-D image using the Geomagic software (Fig. 2 and Supplemental Table
767
S1). Each landmark was represented by ‘x’, ‘y’ and ‘z’ coordinates as part of the Cartesian
768
coordinate system. The coordinates were exported to an Excel spreadsheet for subsequent
769
calculation of 86 Euclidean distances, including 54 linear distances, 10 angular distances and
770
21 indices (ratios) between the linear distances (Fig. 2 and Table 2).
771
Additionally, three direct cranial measurements: maximum cranial breadth (Euryon –
772
Euryon), maximum cranial length (Gonion – Opisthocranium) and maximum cranial height
773
(Vertex – Gnathion), were collected manually using a digital spreading calliper (Paleo-Tech
774
Concepts, USA). Based on the craniofacial and body height measurements, three craniofacial
775
ratios were calculated: Cephalic index: (eu-eu)/(g-op), Head width – Craniofacial height
776
index: (eu-eu)/(v-gn) and Head – Body height index: (v-gn)/(body height), as summarised in
777
Table 2.
778
779
780
781
Table 2. Craniofacial anthropometric measurements recorded in the study and used for genetic
782
association analyses.
Manual craniofacial measurements
•
•
•
V-Gn (Maximum Craniofacial height)
Eu-Eu (Maximum Head Width)
G-Op (Maximum Head Length)
Page 28
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
•
•
•
Cephalic index: (eu-eu)/(g-op)
Head width – Craniofacial height index: (eu-eu)/(v-gn)
Head – Body height index: (v-gn)/(body height)
3D facial measurements
Linear facial distances
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Total face height: tr-gn
Face width: zy-zy
Morphological face height: n-gn
Physiognomical face height: n-sto
Lower profile height: prn-gn
Lower face height: sn-gn
Lower third face depth: t(l)-gn
Middle face depth: t(l)-prn
Middle face height (right): go(r)-zy(r)
Middle face height (left): go(l)-zy(l)
Middle face width 1: t(r)-t(l)
Middle face width 2 (left): zy(l)-al(l)
Middle face width 2 (right): zy(r)-al(r)
Upper face depth: (left): t(l)-tr
Upper face depth: (right): t(r)-tr
Upper third face depth: t(l)-n
Forehead height: g-tr
Extended forehead height: tr-n
Glabella –Gnathion distance: g-gn
Supraorbital depth: t(l)-g
Trichion – Zygion distance (left): tr-zy(l)
Trichion – Zygion distance (right): tr-zy(r)
Nasion - Zygion distance (left): n-zy(l)
Nasion - Zygion distance (right): n-zy(r)
Zygion – Gnathion distance (left): zy(l)-gn
Zygion – Gnathion distance (right): zy(r)-gn
Interendocanthal width: en-en
Interexocanthal width: ex-ex
Eye fissure width (left): en(l)-ex(l)
Eye fissure width (right): en(r)-ex(r)
Eye fissure height (left): ps(l)-pi(l)
Eye fissure height (right): ps(r)-pi(r)
Ear height (left): sa(l)-sba(l)
Ear width (left): t(l)-pa(l)
Nasal bridge length: n-prn
Nose height: n-sn
Nose width: al-al
Nasal tip protrusion: sn-prn
Page 29
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Ala length (left): prn-al(l)
Ala length (right): prn-al(r)
Gonion - Trichion distance (left): go(l)-tr
Gonion - Trichion distance (right): go(r)-tr
Gonion – Glabella distance: g-pg
Pronasale - Gonion distance (left): prn-go(l)
Pronasale - Gonion distance (right): prn-go(r)
Chin height: sl-gn
Mandibular region depth (right): t(r)-gn
Mandible width: go-go
Mandible height: sto-gn
Lower jaw depth (left): gn-go(l)
Lower jaw depth (right): gn-go(r)
Mouth width: ch-ch
Upper vermilion height: ls-sto
Lower vermilion height: li-sto
Angular facial distances
•
•
•
•
•
•
•
•
•
•
Nasal tip angle: (n-prn-sn)
Nasal vertical prominence angle: (tr-prn-gn)
Transverse nasal prominence angle 1: (zy(l)-prn-zy(r))
Transverse nasal prominence angle 2: (t(l)-prn-t(r))
Nasolabial angle: (prn-sn-ls)
Nasofrontal angle: (g-n-prn)
Nasion depth angle: (zy(l)-n-zy(r))
Nasomental angle: (n-prn-pg)
Forehead nasal angle: (tr-n-prn)
Chin prominence angle: (go(l)-gn-go(r))
Ratios (indices)
•
•
•
•
•
•
•
•
•
•
•
•
•
•
Forehead height ratio: (tr-n)/(go(r)-go(l))
Upper face height ratio: (n-sn)/(go(r)-go(l))
Lower face height ratio: (sn-gnx)/(go-go)
Anterior face height 1 ratio: (n-gn)/(go-go)
Anterior face height 2 ratio: (n-gn)/(zy-zy)
Face height index: (n-gn)/(tr-gn)
Upper – Lower face ratio: (tr-g)/(sn-gn)
Upper face height ratio: (n-sn)/(sn-gn)
Upper face width ratio: (n-sn)/(zy-zy)
Total anterior face height ratio: (tr-gn)/(zy-zy)
Mouth width ratio: (ch-ch)x100/(en-en)
Mandible – Face width ratio: (go-go)/(zy-zy)
Mandible index: (sto-gn)x100/(go-go)
Mandible – Interexocanthion distance ratio (go-go)/(ex-ex)
Page 30
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
•
•
•
•
•
•
•
•
Interendocanthion distance ratio: (en-en)/(al-al)
Intercanthal index: (en(r)-en(L)/(ex(r)-ex(l)
Intercanthal – Intracanthal index: (ex(r)-en(r)/(en(l)-ex(l)
Nasal index: (al-al)x100/(n-sn)
Nose-face height index: (n-sn) /(n-gn)
Nose-face width index: (al-al)/(zy-zy)
Nasal tip protrusion – nose width index: (sn-prn)/(al-al)
Nasal tip protrusion –Nose height index: (sn-prn)/(n-sn)
783
784
Phenotypic traits summary
785
A total of 54 linear distances, 10 angular distances and 21 indices (ratios) between the
786
linear distances were calculated based on the Cartesian coordinates of 32 anthropometric
787
landmarks that were manually mapped on each of the 587 3-D facial images (Fig. 2, Fig. 3,
788
Table 2 and Supplemental Table S1). Three additional craniofacial distances were obtained
789
by direct measurement of subjects’ heads and used to calculate three indices: maximum
790
cranial breadth, maximum cranial length and maximum cranial height, cephalic index, head
791
width – craniofacial height index and head – body height index (Table 2). Information on the
792
eyelid and earlobe morphology (single/double and attached/detached respectively) was
793
recorded. Furthermore, the linear and angular facial distances were used to calculate 10
794
principal components (PCs). Additional phenotypic traits such as eye, skin and hair
795
pigmentation, hair texture, freckling, moles, height, weight, BMI, age and sex were collected.
796
In total, the data on 104 craniofacial phenotypic traits were recorded and used for genetic
797
association analyses.
798
The phenotypic data collection by a single examiner achieved more consistent measurements
799
from the 3-D image analyses. In addition, all measurements were based on the images of
800
participants within a narrow age range 26.6 (SD ± 8.9).
801
802
803
DNA extraction and quantification
804
DNA was purified from buccal swabs using the Isohelix DDK isolation kit (Cell Projects,
805
Kent, UK) according to the manufacturer instructions. DNA samples were quantified using a
806
Real Time quantitative PCR (q-PCR) method using a Bio-Rad CFX96 (Bio-Rad, Gladesville,
Page 31
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
807
Australia). This assay amplified a 63bp region of the OCA locus. The primer sequences were
808
5’-GCTGCAGGAGTCAGAAGGTT-3’
809
CATTTGGCGAGCAGAATCC-3’ (reverse primer) at a final concentration of 200mM. All
810
DNA samples were additionally quantified using the Qubit 2.0 fluorimeter (Invitrogen) prior
811
to library construction as per manufacturer recommendations.
(forward
primer)
and
5’-
812
813
Candidate genes and SNPs selection
814
Two main complementary strategies were used to generate a preliminary list of
815
candidate genes and genetic markers. The first focused on searching the literature and web
816
resources for candidate genes involved either in normal craniofacial variation or in
817
craniofacial malformations in humans and model organisms (Supplemental Table S2).
818
The search for candidate genes focused not only on specifically defined craniofacial
819
disorders, but also on genetic syndromes with various manifestations of craniofacial
820
malformations, such as Down syndrome, Noonan Syndrome, Floating-Harbor Syndrome and
821
others, as detailed in Supplemental Table S2. The main resources for locating candidate
822
genes in the animal models were Mouse Genome Informatics [30] and AmiGo tool [31] The
823
main resources for identifying candidate genes in the human genome were OMIM [32] and
824
GeneCards [33]. A comprehensive list of web resources used for candidate gene search is
825
detailed in the Supplemental Appendix S1.
826
The second approach initially implemented a broad search for high Fst SNPs, such as
827
ancestry informative markers (AIMs), with the rationale that many genes affecting
828
craniofacial traits would have significantly different allele frequencies across populations.
829
AIMs were selected from a variety of published and online resources [34-43].
830
The relevant genes obtained by both approaches were subsequently checked for potential
831
involvement in craniofacial embryogenesis, limb development and bilateral body symmetry.
832
It should be noted however, that the final candidate gene list was not limited to craniofacial
833
genes and included high Fst SNPs in genes with unknown function as well as markers located
834
in intergenic regions, potentially possessing regulatory functions.
835
The resulting set of SNPs was further screened for high Fst SNPs (≥0.45) in three ‘1000
836
genomes’ populations (CAU, ASW, CHB) using ENGINES browser [44] as well as
837
potentially functional polymorphisms, such as non-synonymous SNPs [45], markers in
Page 32
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
838
transcription factor binding sites [46] and splicing sites [47] using various web resources, as
839
detailed in Supplemental Appendix S1 and reviewed on the GenEpi website [48]. The
840
candidate markers search resulted in identification of 1,319 SNPs, located in approximately
841
177 genes/intergenic regions, as discussed in the Results section.
842
The chromosomal locations of final candidate markers were submitted to the custom
843
Ampliseq primer design pipeline (Life Technologies), according to manufacturer
844
recommendations. There were primer design difficulties for 881 markers. The marker list was
845
therefore redesigned to include alternative tagging markers showing high linkage
846
disequilibrium with the markers that failed initial primer design, resulting in 1,670 candidate
847
genetic markers. Inclusion of SNPs with MAF<1% added additional 4,381 genetic markers
848
(6,051 in total). The final custom Ampliseq panel was manufactured as two separate pools of
849
849 and 847 primer pairs, with each amplicon covering between 125 bp and 225 bp, therefore
850
possibly containing more than one polymorphism, and in total covering 15.78 kb of the
851
reference human genome. This panel included 1,319 initially targeted craniofacial and
852
pigmentation candidate markers as well as 4,732 markers in LD with original candidate SNPs
853
that failed primer design.
854
Inclusion of novel, rare SNPs (MAF<1%) increased the final number of genotyped markers
855
to 8,518 SNP in all sequenced DNA samples, although the markers with MAF≤2% were not
856
included in the association study. The list of all genotyped markers and their respective genes
857
is detailed in Table S1.
858
SNP genotyping and data analysis
859
Multiple DNA libraries were constructed from sets of 32 Ion XpressTM (Life
860
Technologies) barcoded samples using the Ion AmpliSeqTM library Kit 2.0 (Life
861
Technologies) in conjunction with two custom primer mixes that were pooled according to
862
manufacturer recommendations. Libraries were quantified using the Ion Library Quantitation
863
kit (Life Technologies) and pooled in equal amounts for emulsion PCR, which was
864
performed using the OneTouchTM 2 instrument (Life Technologies) according to
865
manufacturer recommendations. 587 DNA samples were genotyped by massively parallel
866
sequencing on the Personal Genome Machine (PGM) (Life Technologies) using the
867
Sequencing 200 v2 kit and 316 Ion chips (Life Technologies).
868
Raw sequencing data were collected and processed on the Torrent Suite Server v3.6.2 using
869
default settings. Alignment and variant calling were performed against the human genome
Page 33
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
870
reference (hg19) sequence at low stringency settings. Binary alignment map (BAM) files
871
were generated and exported to the Ion ReporterTM (IR) cloud-based software for SNP
872
annotation against the reference hotspot file. The IR analysis resulted in generation of the
873
individual variant caller files (VCF) with genotype calls for each sample as well as various
874
statistics of the sequencing quality.
875
To reduce potential bias of the self-reported ancestry, ancestry inferences were obtained by
876
3,302 markers using STRUCTURE version 2.3.4 with default parameters as per software
877
developer recommendations [49]. SNPs in long-range Linkage Disequilibrium (> 100,000 bp)
878
were excluded from the STRUCTURE run. The ancestry was estimated based on four
879
predefined population clusters: Europeans, East Asians, South Asians and Africans,
880
according to software developer recommendations. Relative allele calls for four predefined
881
HapMap population clusters (CEU, YRI, CHB and JPT) were used as reference populations
882
[50]. The ancestry origin was estimated as a single (unmixed) source where the main ancestry
883
cluster could be affiliated with at least 80% of the total mixed ancestry. The samples with
884
mixed ancestry (>20% admixture) were assigned to an ‘Admixture’ cluster.
885
Association analyses were performed using SNP & Variation Suite v7 (SVS) (Golden Helix,
886
Inc., Bozeman, MT) and replicated using PLINK v1.07 software [51]. Statistical analyses in
887
both software programs were performed using linear regression with quantitative phenotypes,
888
and logistic regressions with binary phenotypes under the assumption of an additive genetic
889
model, while each genotype was numerically encoded as 0, 1 or 2. Population stratification
890
correction, incorporated by EIGENSTRAT program was implemented in the analyses [52,
891
53]. In order to reduce any potential confounding effects, all the craniofacial traits association
892
analyses were performed using sex, BMI and EIGENSTRAT ancestry clusters as covariates.
893
In PLINK, p-values were adjusted using the ‘–adjust’ option. The final reported association
894
results are based on the PLINK statistical analyses with the EIGENSTRAT PCA clusters,
895
BMI and gender as covariates.
896
Annotation analysis of the significantly associated genes was performed using the
897
GeneCards, ENTREZ and UniProtKB web portals [33, 54]. The MalaCards web site was
898
used to detect association between the genes and hereditary syndromes [55]. The GeneMania
899
web site was used to identify a functional network among the genes and encoded proteins
900
[56]. Gene ontology web resource was used to find orthologs of human genes in other
901
organisms [31, 57]. The MGI database was used to search for the phenotype in relevant
Page 34
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
902
craniofacial mouse gene mutants [30]. The dbSNP, 1000 genomes, SNPnexus and Alfred
903
websites were used for SNP annotations [58-61].
904
The SNP Annotation and Proxy Search (SNAP) web portal was used to find SNPs in linkage
905
disequilibrium (LD) and generate LD plots, based on the CEU population panel from the
906
1000 genomes data set, within a distance of up to 500kb and an r2 threshold of 0.8 [62].
907
The Regulome database and potentially functional database (PFS) searches were
908
implemented to annotate SNPs with known and predicted regulatory elements in the
909
intergenic regions of the H. sapiens genome [47, 63].
910
911
List of abbreviations
912
3D: 3-Dimentional; AIMs: ancestry informative markers; ASW: African ancestry in
913
Southwest USA; BAM: Binary alignment map; BMI: Body Mass Index; CAU: Caucasian;
914
CHB: Han Chinese in Beijing, China; ENGINES: ENtire Genome INterface for Exploring
915
Snps; EVT: Externally visible characteristic; DVI: Disaster victim identification; FDP:
916
Forensic DNA phenotyping; GWAS: Genome wide association studies; HWE: Hardy-
917
Weinberg equilibrium; JPT: Japanese in Tokyo, Japan; LD: linkage disequilibrium;
918
lncRNAs: long non-coding RNAs; LINE-1: Long Interspersed Nuclear Element 1; MAF:
919
Minor allele frequency; measurement error; ME: Measurement error; MD: Mean difference;
920
OMIM: Online Mendelian Inheritance in Man; ORFs: open reading frames; pfSNP:
921
Potentially functional SNP; PCA: Principal component analysis; RGB: Red, Green, Blue
922
(colours); SNP: Single-nucleotide polymorphism; SNAP: SNP Annotation and Proxy Search;
923
STR: Short tandem repeat; TF: Transcription factor; VCF: Variant Call Format; YRI: Yoruba
924
in Ibadan, Nigeria.
925
926
Declarations
927
928
Ethics and consent to participate
929
The participants provided their written informed consent to participate in this study, which
930
was approved by the Bond University Ethics committee (RO-510).
Page 35
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
931
932
933
Competing interests
The authors declare that they have no competing interests.
934
935
Authors' contributions
936
MB designed the study, carried out the molecular genetic studies, carried out the data
937
analysis, participated in the statistical analysis and drafted the manuscript. PB performed the
938
statistical analysis and drafted the manuscript. AvD participated in the design of the study
939
and drafted the manuscript. All authors read and approved the final manuscript.
940
941
942
Consent to Publish
Not applicable
943
944
Availability of data and materials
945
The genomic data supporting the conclusions of this article are included within the article and
946
its additional files.
947
Funding
948
The funding for this research was provided by the Technical Support Working Group
949
(Award Number: IS-FB-2946) and Pelerman Holdings Pt Ltd.
950
Acknowledgments
951
We would like to thank the volunteers who participated in this study without whom we could
952
not have performed this research. We would like to thank Olga Kondrashova who helped
Page 36
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
953
with the sample collection. We also thank Technical Support Working Group (TSWG) and
954
Pelerman Holdings Pt Ltd for their generous support of this project.
955
956
References
957
1.
Kohn LAP. The Role of Genetics in Craniofacial Morphology and Growth. Annual
958
Review of Anthropology. 1991;20:261-78. doi: 10.2307/2155802.
959
2.
960
development. 2nd ed. Shelton, CT: People's Medical Pub. House USA; 2010. 250p. ill. (some
961
colour) p.
962
3.
963
growth. Journal of craniofacial genetics and developmental biology. 1986;6(3):289-323.
964
PubMed PMID: 3771738.
965
4.
966
in humans. J Anat. 2009;215(3):240-55. doi: 10.1111/j.1469-7580.2009.01106.x. PubMed
967
PMID: 19531085; PubMed Central PMCID: PMC2750758.
968
5.
969
Genetics. 2009;18(R1):R9-R17. doi: 10.1093/hmg/ddp003.
970
6.
971
Pediatr. 2013;1:53. doi: 10.3389/fped.2013.00053. PubMed PMID: 24400297; PubMed
972
Central PMCID: PMC3873527.
973
7.
974
Craniosynostosis. Semin Pediatr Neurol. 2007;14:150-61.
975
8.
976
Face shape of unaffected parents with cleft affected offspring: combining three-dimensional
977
surface imaging and geometric morphometrics. Orthodontics & Craniofacial Research.
978
2009;12(4):271-81. doi: 10.1111/j.1601-6343.2009.01462.x.
979
9.
980
Anderson, Barry C Powell. Unravelling the molecular control of calvarial suture fusion in
981
children with craniosynostosis. BMC Genomics. 2007;8:458.
982
10.
983
Identification of genes differentially expressed by prematurely fused human sutures using a
Sperber GH, Sperber SM, Guttmann GD. Craniofacial embryogenetics and
Richtsmeier JT, Cheverud JM. Finite element scaling analysis of human craniofacial
Neubauer S, Gunz P, Hublin JJ. The pattern of endocranial ontogenetic shape changes
Sturm RA. Molecular genetics of human pigmentation diversity. Human Molecular
Shkoukani MA, Chen M, Vong A. Cleft Lip - A Comprehensive Review. Front
Kimonis V, Gold J-A, Hoffman TL, Panchal J, Boyadjiev SA. Genetics of
Weinberg SM, Naidoo SD, Bardi KM, Brandon CA, Neiswanger K, Resick JM, et al.
Anna K Coussens CRW, Ian P Hughes, C, Phillip Morris, Angela van Daal, Peter J
Coussens AK, Hughes IP, Wilkinson CR, Morris CP, Anderson PJ, Powell BC, et al.
Page 37
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
984
novel in vivo[thin space]-[thin space]in vitro approach. Differentiation. 2008;76(5):531-45.
985
doi: DOI: 10.1111/j.1432-0436.2007.00244.x.
986
11.
987
Genetic determination of human facial morphology: links between cleft-lips and normal
988
variation. Eur J Hum Genet. 2011;19(11):1192-7. doi: 10.1038/ejhg.2011.110. PubMed
989
PMID: 21694738; PubMed Central PMCID: PMC3198142.
990
12.
991
located in Down syndrome critical region on chromosome 21. DNA research : an
992
international journal for rapid publication of reports on genes and genomes. 1997;4(5):321-4.
993
doi: 10.1093/dnares/4.5.321. PubMed PMID: 9455479.
994
13.
995
abolish DNA binding in Saethre-Chotzen syndrome. FEBS Lett. 2001;492:112-8.
996
14.
997
Craniosynostosis in Alagille syndrome. Am J Med Genet. 2002;112(2):176-80. doi:
998
10.1002/ajmg.10608. PubMed PMID: 12244552.
999
15.
Boehringer S, van der Lijn F, Liu F, Gunther M, Sinigerova S, Nowak S, et al.
Nakamura A, Hattori M, Sakaki Y. A novel gene isolated from human placenta
El Ghouzzi V. Mutations in the basic domain and the loop-helix II junction of TWIST
Kamath BM, Stolle C, Bason L, Colliton RP, Piccoli DA, Spinner NB, et al.
Hood RL, Lines MA, Nikkel SM, Schwartzentruber J, Beaulieu C, Nowaczyk MJ, et
1000
al. Mutations in SRCAP, encoding SNF2-related CREBBP activator protein, cause Floating-
1001
Harbor syndrome. Am J Hum Genet. 2012;90(2):308-13. doi: 10.1016/j.ajhg.2011.12.001.
1002
PubMed PMID: 22265015; PubMed Central PMCID: PMC3276662.
1003
16.
1004
cerebellar response to mitogenic Hedgehog signaling in Down's syndrome mice. Proc Natl
1005
Acad Sci U S A. 2006;103(5):1452-6.
1006
17.
1007
Noonan syndrome PTPN11 gene mutation--phenotype characterization. American journal of
1008
medical genetics Part A. 2008;146A(3):350-3. doi: 10.1002/ajmg.a.32140. PubMed PMID:
1009
18203203.
1010
18.
1011
reliability using different three-dimensional scanning systems. Forensic Sci Int. 2011;207(1-
1012
3):127-34. doi: 10.1016/j.forsciint.2010.09.018. PubMed PMID: 20951517.
1013
19.
1014
tissue landmarks on 3D laser-scanned facial images. Orthodontics & Craniofacial Research.
1015
2009;12(1):33-42. doi: 10.1111/j.1601-6343.2008.01435.x.
1016
20.
1017
Papadopulos NA, et al. Accuracy and Precision of the Three-Dimensional Assessment of the
Roper RJ, Baxter LL, Saran NG, Klinedinst DK, Beachy PA, Reeves RH. Defective
Croonen EA, van der Burgt I, Kapusta L, Draaisma JM. Electrocardiography in
Fourie Z, Damstra J, Gerrits PO, Ren Y. Evaluation of anthropometric accuracy and
Toma AM, Zhurov A, Playle R, Ong E, Richmond S. Reproducibility of facial soft
Kovacs L, Zimmermann A, Brockmann G, Baurecht H, Schwenzer-Zimmerer K,
Page 38
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1018
Facial Surface Using a 3-D Laser Scanner. IEEE Transactions on Medical Imaging.
1019
2006;25(6):742-54. PubMed PMID: 21197761.
1020
21.
1021
haplotype-tag SNP associated with normal variation in craniofacial shape. Genomics.
1022
2005;85(5):563-73. doi: DOI: 10.1016/j.ygeno.2005.02.002.
1023
22.
1024
Genome-wide association study of three-dimensional facial morphology identifies a variant
1025
in PAX3 associated with nasion position. Am J Hum Genet. 2012;90(3):478-85. Epub
1026
2012/02/22. doi: 10.1016/j.ajhg.2011.12.021. PubMed PMID: 22341974; PubMed Central
1027
PMCID: PMC3309180.
1028
23.
1029
genome-wide association study identifies five loci influencing facial morphology in
1030
Europeans. PLoS Genet. 2012;8(9):e1002932. doi: 10.1371/journal.pgen.1002932. PubMed
1031
PMID: 23028347; PubMed Central PMCID: PMC3441666.
1032
24.
1033
candidate gene and GWAS Approaches in Asthma. PLoS One. 2010;5(11):e13894. doi:
1034
10.1371/journal.pone.0013894. PubMed PMID: 21103062; PubMed Central PMCID:
1035
PMC2980484.
1036
25.
1037
dimensional evidence-based candidate gene prioritization approach for complex diseases-
1038
schizophrenia
1039
10.1093/bioinformatics/btp428. PubMed PMID: 19602527; PubMed Central PMCID:
1040
PMC2752609.
1041
26.
1042
Nubia). Am J Phys Anthropol. 1978;49(2):277-82. doi: 10.1002/ajpa.1330490217. PubMed
1043
PMID: 717558.
1044
27.
1045
Archives of Dermatology. 1988;124(6):869.
1046
28.
1047
Melanoma Res. 2009;22(5):544-62. doi: 10.1111/j.1755-148X.2009.00606.x. PubMed
1048
PMID: 19619260.
1049
29.
1050
1994. xix, 405 p. p.
Coussens AK, Daal Av. Linkage disequilibrium analysis identifies an FGFR1
Paternoster L, Zhurov AI, Toma AM, Kemp JP, St Pourcain B, Timpson NJ, et al.
Liu F, van der Lijn F, Schurmann C, Zhu G, Chakravarty MM, Hysi PG, et al. A
Michel S, Liang L, Depner M, Klopp N, Ruether A, Kumar A, et al. Unifying
Sun J, Jia P, Fanous AH, Webb BT, van den Oord EJ, Chen X, et al. A multi-
as
a
case.
Bioinformatics.
2009;25(19):2595-6602.
doi:
Hrdy DB. Analysis of hair samples of mummies from Semma South (Sudanese
Fitzpatrick TB. The validity and practicality of sun-reactive skin types I through VI.
Sturm RA, Larsson M. Genetics of human iris colour and patterns. Pigment Cell
Farkas LG. Anthropometry of the head and face. 2nd ed. New York: Raven Press;
Page 39
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1051
30.
Site MMRW. The Jackson Laboratory, Bar Harbor, Maine. World Wide Web
1052
(http://mousemutant.jax.org/) Bar Harbor, Maine.2010 [cited 2013 March]. Available from:
1053
http://mousemutant.jax.org/.
1054
31.
1055
ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet.
1056
2000;25(1):25-9. doi: 10.1038/75556. PubMed PMID: 10802651; PubMed Central PMCID:
1057
PMC3037419.
1058
32.
1059
Medicine, Johns Hopkins University (Baltimore, MD) [cited 2012 March]. Available from:
1060
http://omim.org/.
1061
33.
1062
about genes, proteins and diseases. Trends in Genetics. 1997;13(4):163-. doi: Doi
1063
10.1016/S0168-9525(97)01103-7. PubMed PMID: WOS:A1997WQ96900011.
1064
34.
1065
wide detection and characterization of positive selection in human populations. Nature.
1066
2007;449(7164):913-8. doi: 10.1038/nature06250. PubMed PMID: 17943131; PubMed
1067
Central PMCID: PMC2687721.
1068
35.
1069
HapMap
1070
doi:10.1186/1471-2105-8-484.
1071
36.
1072
informative marker sets for determining continental origin and admixture proportions in
1073
common populations in America. Hum Mutat. 2009;30(1):69-78. doi: 10.1002/humu.20822.
1074
PubMed PMID: 18683858; PubMed Central PMCID: PMC3073397.
1075
37.
1076
individual assignment to worldwide populations. Journal of Medical Genetics. doi:
1077
10.1136/jmg.2010.078212.
1078
38.
1079
a cost-effective panel of ancestry informative markers for determining continental origins.
1080
PLoS One. 2010;5(10):e13443. doi: 10.1371/journal.pone.0013443. PubMed PMID:
1081
20976178; PubMed Central PMCID: PMC2955551.
1082
39.
1083
multiplex for the simultaneous prediction of biogeographic ancestry and pigmentation type.
1084
Forensic Sci Inter: Genet Suppl. 2011;3(1):e500-e1. doi: 10.1016/j.fsigss.2011.10.001.
Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene
Online Mendelian Inheritance in Man O. McKusick-Nathans Institute of Genetic
Rebhan M, ChalifaCaspi V, Prilusky J, Lancet D. GeneCards: Integrating information
Sabeti PC, Varilly P, Fry B, Lohmueller J, Hostetter E, Cotsapas C, et al. Genome-
Zhou N, Wang L. Effective selection of informative SNPs and classification on the
genotype
data.
BMC
Bioinformatics.
2007;8(1):484.
PubMed
PMID:
Kosoy R, Nassir R, Tian C, White PA, Butler LM, Silva G, et al. Ancestry
Paschou P, Lewis J, Javed A, Drineas P. Ancestry informative markers for fine-scale
Londin ER, Keller MA, Maista C, Smith G, Mamounas LA, Zhang R, et al. CoAIMs:
Bulbul O, Filoglu G, Altuncul H, Aradas AF, Ruiz Y, Fondevila M, et al. A SNP
Page 40
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1085
40.
Tandon A, Patterson N, Reich D. Ancestry informative marker panels for African
1086
Americans based on subsets of commercially available SNP arrays. Genet Epidemiol.
1087
2011;35(1):80-3. doi: 10.1002/gepi.20550. PubMed PMID: 21181899.
1088
41.
1089
Eurasiaplex: a forensic SNP assay for differentiating European and South Asian ancestries.
1090
Forensic Sci Int Genet. 2013;7(3):359-66. doi: 10.1016/j.fsigen.2013.02.010. PubMed PMID:
1091
23537756.
1092
42.
1093
50-SNP assay for biogeographic ancestry and phenotype prediction in the U.S. population.
1094
Forensic Sci Int Genet. 2014;8(1):101-8. doi: 10.1016/j.fsigen.2013.07.010. PubMed PMID:
1095
24315596.
1096
43.
1097
toward an efficient panel of SNPs for ancestry inference. Forensic Sci Int Genet.
1098
2014;10C(0):23-32. doi: 10.1016/j.fsigen.2014.01.002. PubMed PMID: 24508742.
1099
44.
1100
entire human genomes. BMC Bioinformatics. 12(1):105. PubMed PMID: doi:10.1186/1471-
1101
2105-12-105.
1102
45.
1103
method and server for predicting damaging missense mutations. Nat Methods. 2010;7(4):248-
1104
9. doi: 10.1038/nmeth0410-248. PubMed PMID: 20354512; PubMed Central PMCID:
1105
PMC2855889.
1106
46.
1107
identification of putative transcription factor binding sites in multiple genomes. BMC
1108
Bioinformatics. 2005;6(1):79. PubMed PMID: doi:10.1186/1471-2105-6-79.
1109
47.
1110
functional SNP resource that facilitates hypotheses generation through knowledge syntheses.
1111
Human Mutation. 32(1):19-24. doi: 10.1002/humu.21331.
1112
48.
1113
constantly updated survival guide for genetic epidemiology. The GenEpi Toolbox.
1114
Atherosclerosis. 209(2):321-35. doi: DOI: 10.1016/j.atherosclerosis.2009.10.026.
1115
49.
1116
multilocus
1117
WOS:000087475100039.
Phillips C, Freire Aradas A, Kriegel AK, Fondevila M, Bulbul O, Santos C, et al.
Gettings KB, Lai R, Johnson JL, Peck MA, Hart JA, Gordish-Dressman H, et al. A
Kidd KK, Speed WC, Pakstis AJ, Furtado MR, Fang R, Madbouly A, et al. Progress
Amigo J, Salas A, Phillips C. ENGINES: exploring single nucleotide variation in
Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, et al. A
Marinescu V, Kohane I, Riva A. MAPPER: a search engine for the computational
Wang J, Ronaghi M, Chong SS, Lee CGL. pfSNP: An integrated potentially
Coassin S, Brandstätter A, Kronenberg F. Lost in the space of bioinformatic tools: A
Pritchard JK, Stephens M, Donnelly P. Inference of population structure using
genotype
data.
Genetics.
2000;155(2):945-59.
PubMed
PMID:
Page 41
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1118
50.
Thorisson G, Smith A, Krishnan L, Stein L. The International HapMap Project Web
1119
site. Genome Res. 2005;15:1592.
1120
51.
1121
a tool set for whole-genome association and population-based linkage analyses. Am J Hum
1122
Genet. 2007;81(3):559-75. doi: 10.1086/519795. PubMed PMID: 17701901; PubMed Central
1123
PMCID: PMC1950838.
1124
52.
1125
components analysis corrects for stratification in genome-wide association studies. Nat
1126
Genet. 2006;38(8):904-9. doi: 10.1038/ng1847. PubMed PMID: 16862161.
1127
53.
1128
2006;2(12):e190. doi: 10.1371/journal.pgen.0020190. PubMed PMID: 17194218; PubMed
1129
Central PMCID: PMC1713260.
1130
54.
1131
2014;42(Database issue):D191-8. doi: 10.1093/nar/gkt1140. PubMed PMID: 24253303;
1132
PubMed Central PMCID: PMC3965022.
1133
55.
1134
an integrated compendium for diseases and their annotation. Database : the journal of
1135
biological databases and curation. 2013;2013:bat018. doi: 10.1093/database/bat018. PubMed
1136
PMID: 23584832; PubMed Central PMCID: PMC3625956.
1137
56.
1138
GeneMANIA prediction server: biological network integration for gene prioritization and
1139
predicting gene function. Nucleic Acids Res. 2010;38(Web Server issue):W214-20. doi:
1140
10.1093/nar/gkq537. PubMed PMID: 20576703; PubMed Central PMCID: PMC2896186.
1141
57.
1142
Genome Project: a unified framework for functional annotation across species. PLoS Comput
1143
Biol. 2009;5(7):e1000431. doi: 10.1371/journal.pcbi.1000431. PubMed PMID: 19578431;
1144
PubMed Central PMCID: PMC2699109.
1145
58.
1146
the NCBI database of genetic variation. Nucleic Acids Research. 2001;29(1):308-11. doi:
1147
10.1093/nar/29.1.308.
1148
59.
1149
map of human genome variation from population-scale sequencing. Nature. 467(7319):1061 -
1150
73. PubMed PMID: doi:10.1038/nature09534.
Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, Bender D, et al. PLINK:
Price AL, Patterson NJ, Plenge RM, Weinblatt ME, Shadick NA, Reich D. Principal
Patterson N, Price AL, Reich D. Population structure and eigenanalysis. PLoS Genet.
UniProt C. Activities at the Universal Protein Resource (UniProt). Nucleic Acids Res.
Rappaport N, Nativ N, Stelzer G, Twik M, Guan-Golan Y, Stein TI, et al. MalaCards:
Warde-Farley D, Donaldson SL, Comes O, Zuberi K, Badrawi R, Chao P, et al. The
Reference Genome Group of the Gene Ontology C. The Gene Ontology's Reference
Sherry ST, Ward MH, Kholodov M, Baker J, Phan L, Smigielski EM, et al. dbSNP:
Durbin RM, Abecasis GR, Altshuler DL, Auton A, Brooks LD, Gibbs RA, et al. A
Page 42
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1151
60.
Chelala C, Khan A, Lemoine NR. SNPnexus: a web database for functional
1152
annotation of newly discovered and public domain single nucleotide polymorphisms.
1153
Bioinformatics. 2009;25(5):655-61. doi: 10.1093/bioinformatics/btn653. PubMed PMID:
1154
19098027; PubMed Central PMCID: PMC2647830.
1155
61.
1156
ALFRED: an allele frequency database for microevolutionary studies. Evol Bioinform
1157
Online. 2005;1:1-10. PubMed PMID: 19325849; PubMed Central PMCID: PMC2658869.
1158
62.
1159
SNAP: a web-based tool for identification and annotation of proxy SNPs using HapMap.
1160
Bioinformatics. 2008;24(24):2938-9. doi: 10.1093/bioinformatics/btn564. PubMed PMID:
1161
18974171; PubMed Central PMCID: PMC2720775.
1162
63.
1163
Annotation of functional variation in personal genomes using RegulomeDB. Genome Res.
1164
2012;22(9):1790-7. doi: 10.1101/gr.137323.112. PubMed PMID: 22955989; PubMed Central
1165
PMCID: PMC3431494.
1166
64.
1167
tool and its accuracy compared with direct facial anthropometric measurements. British
1168
journal of plastic surgery. 1995;48(8):551-8. PubMed PMID: 8548155.
1169
65.
1170
Evaluation of scanning parameters for a surface colour laser scanner. International Congress
1171
Series. 2004;1268:1162-7. doi: 10.1016/j.ics.2004.03.264. PubMed PMID: 13327565.
1172
66.
1173
applications. American journal of orthodontics and dentofacial orthopedics : official
1174
publication of the American Association of Orthodontists, its constituent societies, and the
1175
American Board of Orthodontics. 2002;122(4):342-8. PubMed PMID: 12411877.
1176
67.
1177
structured light techniques. Computer Methods & Programs in Biomedicine. 2009;94(3):290-
1178
8. doi: 10.1016/j.cmpb.2009.01.010. PubMed PMID: 37572488.
1179
68.
1180
three-dimensional facial scans. European journal of orthodontics. 2006;28(5):408-15. doi:
1181
10.1093/ejo/cjl024. PubMed PMID: 16901962.
1182
69.
1183
improves functional interpretation of cis-regulatory regions. Nat Biotechnol. 2010;28(5):495-
1184
501. doi: 10.1038/nbt.1630. PubMed PMID: 20436461.
Rajeevan H, Cheung KH, Gadagkar R, Stein S, Soundararajan U, Kidd JR, et al.
Johnson AD, Handsaker RE, Pulit SL, Nizzari MM, O'Donnell CJ, de Bakker PI.
Boyle AP, Hong EL, Hariharan M, Cheng Y, Schaub MA, Kasowski M, et al.
Aung SC, Ngim RC, Lee ST. Evaluation of the laser scanner as a surface measuring
Bianchi SD, Spada MC, Bianchi L, Verzè L, Vezzetti E, Tornincasa S, et al.
Kusnoto B, Evans CA. Reliability of a 3D surface laser scanner for orthodontic
Ma L, Xu T, Lin J. Validation of a three-dimensional facial scanning system based on
Gwilliam JR, Cunningham SJ, Hutton T. Reproducibility of soft tissue landmarks on
McLean CY, Bristor D, Hiller M, Clarke SL, Schaar BT, Lowe CB, et al. GREAT
Page 43
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1185
70.
Weston EM, Friday AE, Lio P. Biometric evidence that sexual selection has shaped
1186
the hominin face. PLoS One. 2007;2(8):e710. doi: 10.1371/journal.pone.0000710. PubMed
1187
PMID: 17684556; PubMed Central PMCID: PMC1937021.
1188
71.
1189
multiethnic samples. Hum Mol Genet. 2008;17(R2):R151-5. Epub 2008/10/15. doi:
1190
10.1093/hmg/ddn263. PubMed PMID: 18852204; PubMed Central PMCID: PMC2782359.
1191
72.
1192
stratification via individual ancestry estimates versus self-reported race. Cancer Epidemiol
1193
Biomarkers Prev. 2005;14(6):1545-51. doi: 10.1158/1055-9965.EPI-04-0832. PubMed
1194
PMID: 15941970.
1195
73.
1196
structure, self-identified race/ethnicity, and confounding in case-control association studies.
1197
Am J Hum Genet. 2005;76(2):268-75. doi: 10.1086/427888. PubMed PMID: 15625622;
1198
PubMed Central PMCID: PMC1196372.
1199
74.
1200
stratification in genome-wide association studies. Nat Rev Genet. 2010;11(7):459-63. Epub
1201
2010/06/16. doi: 10.1038/nrg2813. PubMed PMID: 20548291; PubMed Central PMCID:
1202
PMC2975875.
1203
75.
1204
genome-wide significance threshold be? Empirical replication of borderline genetic
1205
associations.
1206
10.1093/ije/dyr178. PubMed PMID: 22253303.
1207
76.
1208
association scans. Genet Epidemiol. 2008;32(3):227-34. doi: 10.1002/gepi.20297. PubMed
1209
PMID: 18300295; PubMed Central PMCID: PMC2573032.
1210
77.
1211
craniofacial morphology by distant-acting enhancers. Science. 2013;342(6157):1241006. doi:
1212
10.1126/science.1241006. PubMed PMID: 24159046.
1213
78.
1214
PANTHER: a library of protein families and subfamilies indexed by function. Genome Res.
1215
2003;13(9):2129-41. doi: 10.1101/gr.772403. PubMed PMID: 12952881; PubMed Central
1216
PMCID: PMC403709.
1217
79.
Cooper RS, Tayo B, Zhu X. Genome-wide association studies: implications for
Barnholtz-Sloan JS, Chakraborty R, Sellers TA, Schwartz AG. Examining population
Tang H, Quertermous T, Rodriguez B, Kardia SL, Zhu X, Brown A, et al. Genetic
Price AL, Zaitlen NA, Reich D, Patterson N. New approaches to population
Panagiotou OA, Ioannidis JP, Genome-Wide Significance P. What should the
International
journal
of
epidemiology.
2012;41(1):273-86.
doi:
Dudbridge F, Gusnanto A. Estimation of significance thresholds for genomewide
Attanasio C, Nord AS, Zhu Y, Blow MJ, Li Z, Liberton DK, et al. Fine tuning of
Thomas PD, Campbell MJ, Kejariwal A, Mi H, Karlak B, Daverman R, et al.
Goymer P. Synonymous mutations break their silence. Nat Rev Genet. 2007;8(2):92-.
Page 44
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1218
80.
Zhao Y, Guo YJ, Tomac AC, Taylor NR, Grinberg A, Lee EJ, et al. Isolated cleft
1219
palate in mice with a targeted mutation of the LIM homeobox gene lhx8. Proc Natl Acad Sci
1220
U S A. 1999;96(26):15002-6. PubMed PMID: 10611327; PubMed Central PMCID:
1221
PMC24762.
1222
81.
1223
expression patterns of two LIM-homeodomain genes, Lhx6 and L3/Lhx8, in the developing
1224
palate. Orthod Craniofac Res. 2002;5(2):65-70. PubMed PMID: 12086327.
1225
82.
1226
new gene, EVC2, is mutated in Ellis–van Creveld syndrome. Molecular genetics and
1227
metabolism. 2002;77(4):291-5. doi: 10.1016/s1096-7192(02)00178-6.
1228
83.
1229
al. Mandibuloacral dysplasia is caused by a mutation in LMNA-encoding lamin A/C. Am J
1230
Hum Genet. 2002;71(2):426-31. doi: 10.1086/341908. PubMed PMID: 12075506; PubMed
1231
Central PMCID: PMC379176.
1232
84.
1233
homologue of the Drosophila eyes absent gene underlies branchio-oto-renal (BOR) syndrome
1234
and identifies a novel gene family. Nat Genet. 1997;15(2):157-64. doi: 10.1038/ng0297-157.
1235
PubMed PMID: 9020840.
1236
85.
1237
independent function of gene and pseudogene mRNAs regulates tumour biology. Nature.
1238
2010;465(7301):1033-8.
1239
86.
1240
expressed pseudogene regulates the messenger-RNA stability of its homologous coding gene.
1241
Nature. 2003;423(6935):91-6.
1242
87.
1243
gene CELSR1 in neural tube defects and caudal agenesis. Birth defects research Part A,
1244
Clinical and molecular teratology. 2012;94(3):176-81. doi: 10.1002/bdra.23002. PubMed
1245
PMID: 22371354.
1246
88.
1247
components in embryonic eyelid epithelium. PLoS One. 2014;9(2):e87038. doi:
1248
10.1371/journal.pone.0087038. PubMed PMID: 24498290; PubMed Central PMCID:
1249
PMC3911929.
Zhang Y, Mori T, Takaki H, Takeuch M, Iseki K, Hagino S, et al. Comparison of the
Galdzicka M, Patnala S, Hirshman MG, Cai JF, Nitowsky H, A Egeland J, et al. A
Novelli G, Muchir A, Sangiuolo F, Helbling-Leclerc A, D'Apice MR, Massart C, et
Abdelhak S, Kalatzis V, Heilig R, Compain S, Samson D, Vincent C, et al. A human
Poliseno L, Salmena L, Zhang J, Carver B, Haveman WJ, Pandolfi PP. A coding-
Hirotsune S, Yoshida N, Chen A, Garrett L, Sugiyama F, Takahashi S, et al. An
Allache R, De Marco P, Merello E, Capra V, Kibar Z. Role of the planar cell polarity
Meng Q, Jin C, Chen Y, Chen J, Medvedovic M, Xia Y. Expression of signaling
Page 45
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1250
89.
Visscher PM. Sizing up human height variation. Nature Genetics. 2008;40(5):489-90.
1251
doi: 10.1038/ng0508-489. PubMed PMID: 18443579.
1252
90.
1253
role of common variation in the genomic and biological architecture of adult human height.
1254
Nat Genet. 2014;46(11):1173-86. doi: 10.1038/ng.3097. PubMed PMID: 25282103; PubMed
1255
Central PMCID: PMCPMC4250049.
1256
91. Claes P, Liberton DK, Daniels K, Rosana KM, Quillen EE, Pearson LN, et al. Modeling
1257
3D
1258
10.1371/journal.pgen.1004224. PubMed PMID: 24651127; PubMed Central PMCID:
1259
PMC3961191.
1260
92. Lau E. Non-coding RNA: Zooming in on lncRNA functions. Nat Rev Genet.
1261
2014;15(9):574-5. doi: 10.1038/nrg3795.
1262
93. Wu H, Nord AS, Akiyama JA, Shoukry M, Afzal V, Rubin EM, et al. Tissue-Specific
1263
RNA
1264
2014;10(9):e1004610. doi: 10.1371/journal.pgen.1004610.
1265
94.
1266
Common Human Facial Morphological Variation Using High Density 3D Image
1267
Registration. PLoS Comput Biol. 2013;9(12):e1003375. doi: 10.1371/journal.pcbi.100337.
1268
95.
1269
a phenotype. Computer Vision and Pattern Recognition, 2008 CVPR 2008 IEEE Conference
1270
on; 2008: IEEE.
1271
96.
1272
results
1273
10.1016/j.fsigen.2014.08.008. PubMed PMID: 25194685.
Wood AR, Esko T, Yang J, Vedantam S, Pers TH, Gustafsson S, et al. Defining the
facial
shape
Expression
from
Marks
DNA.
PLoS
Distant-Acting
Genet.
2014;10(3):e1004224.
Developmental
Enhancers.
PLoS
doi:
Genet.
Peng S, Tan J, Hu S, Zhou H, Guo J, Jin L, et al. Detecting Genetic Association of
Wolf L, Donner Y, editors. An experimental study of employing visual appearance as
Claes P, Hill H, Shriver MD. Toward DNA-based facial composites: preliminary
and
validation.
Forensic
Sci
Int
Genet.
2014;13:208-16.
doi:
Page 46
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1274
Figure legends and additional file descriptions
1275
1276
Figure 1. Anatomical position of the 32 manually annotated anthropometric landmarks
1277
used for calculation of linear and angular distances and ratios between the linear
1278
distances. Some landmarks are not clearly visible due to image orientation. gn = Gnathion,
1279
pg= Pogonion, sl = Sublabiale, li = Labiale Inferius, sto = Stomion, ls = Labiale superius, ch-
1280
r = Chelion right, ch-l = Chelion left, go-r = Gonion Right, go-l = Gonion left, sn =
1281
Subnasale, prn= Pronasale, al-r = Alare right; al-l= Alare left, n = Nasion, g= Glabella; tr =
1282
Tragion, en-l = left Endocanthion, en-r = right Endocanthion, ex-r = Right Endocanthion; ex-l
1283
= left Endocanthion, ps-r = Palpebrale superius right, ps-l = Palpebrale superius left , pi-r =
1284
Palpebrale inferius right, pi-l = Palpebrale inferius left, zy-r = Zygion Right, zy-l = Zygion
1285
Left, pra-r = Tragion right, pra-l = Tragion Left, sba-l = Subalare left, sa-l = Superaurale Left,
1286
pa-l = Postaurale left.
1287
1288
Figure 2. Illustration of linear and angular distances calculated from manually
1289
annotated landmark coordinates.
1290
1291
Figure 3. Population structure as represented by plotting genomic PCs 1 and 2, using
1292
270 HapMap individuals as anchor clusters. YRI: Yoruba, Nigeria, Africa. JRI: Japanese,
1293
Tokyo, Japan. CHB: Han Chinese, Beijing, China. CEU: Utah residents with European
1294
ancestry.
Page 47
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1295
Additional files
1296
Figure S1. Q-Q plot of the PCA-corrected −log10 p-values for the
1297
difference between the observed association for the tails of al-al distance
1298
and expected association based on the overall al-al distance distribution.
1299
Figure S2. Q-Q plot of the PCA-corrected −log10 p-values for the
1300
difference between the observed association for the tails of sn-prn distance
1301
and expected association based on the overall sn-prn distance distribution.
1302
Figure S3. Q-Q plot of the PCA-corrected −log10 p-values for the
1303
difference between the observed association for the tails of cephalic index
1304
and expected association based on the overall cephalic index distribution.
1305
Figure S4. Q-Q plot of the PCA-corrected −log10 p-values for the
1306
difference between the observed association for the tails of nasal index and
1307
expected association based on the overall nasal index distribution.
1308
Figure S5. Q-Q plot of the PCA-corrected −log10 p-values for the
1309
difference between the observed association for the tails of nose-face width
1310
index and expected association based on the overall nose-face width index
1311
distribution.
1312
Figure S6. Q-Q plot of the PCA-corrected −log10 p-values for the
1313
difference between the observed association for the tails of nasolabial angle
1314
and expected association based on the overall nasolabial angle distance
1315
distribution.
1316
Page 48
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1317
Figure S7. Q-Q plot of the PCA-corrected −log10 p-values for the
1318
difference between the observed association for the tails of transverse nasal
1319
prominence angle and expected association based on the overall transverse
1320
nasal prominence angle distribution.
1321
Figure S8. Q-Q plot of the PCA-corrected −log10 p-values for the
1322
difference between the observed association for the tails of PC1 trait and
1323
expected association based on the overall PC1 trait distance distribution.
1324
Figure S9. Manhattan plot of the genomic associations of the al-al distance,
1325
based on the initial p-values from analysis of the PCA-corrected data. The
1326
−log10 (P value) is plotted against the physical positions of each SNP on
1327
each chromosome. The basic significance threshold is indicated by the blue
1328
line for -log10(1e-5) and the genome-wide significance threshold for -
1329
log10(5e-8) is indicated by the red line.
1330
Figure S10. Manhattan plot of the genomic associations of the sn-prn
1331
distance, based on the initial p-values from analysis of the PCA-corrected
1332
data. The −log10 (P value) is plotted against the physical positions of each
1333
SNP on each chromosome. The basic significance threshold is indicated by
1334
the blue line for -log10(1e-5) and the genome-wide significance threshold
1335
for -log10(5e-8) is indicated by the red line.
1336
Figure S11. Manhattan plot of the genomic associations of the cephalic
1337
index, based on the initial p-values from analysis of the PCA-corrected
1338
data. The −log10 (P value) is plotted against the physical positions of each
1339
SNP on each chromosome. The basic significance threshold is indicated by
1340
the blue line for -log10(1e-5) and the genome-wide significance threshold
1341
for -log10(5e-8) is indicated by the red line.
1342
Figure S12. Manhattan plot of the genomic associations of the nasal index,
1343
based on the initial p-values from analysis of the PCA-corrected data. The
Page 49
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1344
−log10 (P value) is plotted against the physical positions of each SNP on
1345
each chromosome. The basic significance threshold is indicated by the blue
1346
line for -log10(1e-5) and the genome-wide significance threshold for -
1347
log10(5e-8) is indicated by the red line.
1348
Figure S13. Manhattan plot of the genomic associations of the nose-face
1349
width index, based on the initial p-values from analysis of the PCA-
1350
corrected data. The −log10 (P value) is plotted against the physical
1351
positions of each SNP on each chromosome. The basic significance
1352
threshold is indicated by the blue line for -log10(1e-5) and the genome-wide
1353
significance threshold for -log10(5e-8) is indicated by the red line.
1354
Figure S14. Manhattan plot of the genomic associations of the nasolabial
1355
angle, based on the initial p-values from analysis of the PCA-corrected
1356
data. The −log10 (P value) is plotted against the physical positions of each
1357
SNP on each chromosome. The basic significance threshold is indicated by
1358
the blue line for -log10(1e-5) and the genome-wide significance threshold
1359
for -log10(5e-8) is indicated by the red line.
1360
Figure S15. Manhattan plot of the genomic associations of the transverse
1361
nasal prominence angle, based on the initial p-values from analysis of the
1362
PCA-corrected data. The −log10 (P value) is plotted against the physical
1363
positions of each SNP on each chromosome. The basic significance
1364
threshold is indicated by the blue line for -log10(1e-5) and the genome-wide
1365
significance threshold for -log10(5e-8) is indicated by the red line.
1366
Figure S16. Manhattan plot of the genomic associations of the PC1 trait,
1367
based on the initial p-values from analysis of the PCA-corrected data. The
1368
−log10 (P value) is plotted against the physical positions of each SNP on
1369
each chromosome. The basic significance threshold is indicated by the blue
1370
line for -log10(1e-5) and the genome-wide significance threshold for -
1371
log10(5e-8) is indicated by the red line.
Page 50
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1372
Figure S17. Pie chart, illustrating molecular function classification of
1373
human genes, harbouring genomic markers in significant association with
1374
craniofacial phenotypes. The genes include: AGXT2, BMP4, CACNB4,
1375
COL11A1, EGFR, EYA1, EYA2, FAM49A, FOXN3, LMNA, MYO5A, PAX3,
1376
PCDH15, RTTN, SMAD1, XXYLT1 and ZEB1.
1377
Figure S18. Pie chart, illustrating biological processes classification
1378
involving human genes, harbouring genomic markers in significant
1379
association with craniofacial phenotypes. The genes include: AGXT2, BMP4,
1380
CACNB4, COL11A1, EGFR, EYA1, EYA2, FAM49A, FOXN3, LMNA, MYO5A,
1381
PAX3, PCDH15, RTTN, SMAD1, XXYLT1 and ZEB1.
1382
Figure S19. Pie chart, illustrating protein product classification of the
1383
human genes, harbouring genomic markers in significant association with
1384
craniofacial phenotypes. The genes are: AGXT2, BMP4, CACNB4, COL11A1,
1385
EGFR, EYA1, EYA2, FAM49A, FOXN3, LMNA, MYO5A, PAX3, PCDH15,
1386
RTTN, SMAD1, XXYLT1 and ZEB1.
1387
Table S1. Manually annotated facial landmarks used in the study.
1388
Table S2. Genetic associations with pigmentation traits. Gene: gene name;
1389
rs#: reference SNP ID number; SNP: chromosomal location of the marker;
1390
Genomic annotation: genomic location of the marker; UNADJ: Unadjusted p-
1391
values; BONF: Bonferroni single-step adjusted; HOLM: Holm (1979) step-
1392
down adjusted; SIDAK_SS: Sidak single-step adjusted; SIDAK_SD: Sidak
1393
step-down adjusted; FDR_BH: Benjamini & Hochberg (1995) step-up FDR
1394
control; FDR_BY: Benjamini & Yekutieli (2001) step-up FDR control.
1395
Table S3. Genetic syndromes displaying various craniofacial abnormalities,
1396
used to locate candidate genes for the study.
1397
Appendix S1. Comprehensive list of web resources used for candidate gene
1398
search and its output. Note the presence of multiple tabs in this spreadsheet.
Page 51
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
1399
Appendix S2. Three spreadsheets, detailing a list of 8,518 genetic markers
1400
genotyped in 587 DNA samples and a list of 2,332 markers used for
1401
association analyses, following MAF (2%) and HWE filtering.
Page 52
bioRxiv preprint first posted online Jul. 20, 2016; doi: http://dx.doi.org/10.1101/060814. The copyright holder for this preprint (which was not
peer-reviewed) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Page 53